the mice at play in the califa surveyvw8/resources/mice_v8_astroph.pdf ·...

Astronomy& Astrophysicsmanuscript no. mice_v8_astroph c©ESO 2014May 30, 2014

The Mice at play in the CALIFA survey

A case study of a gas-rich major merger between first passage and coalescence

Vivienne Wild1,2,⋆, Fabian Rosales-Ortega3,4, Jesus Falcón-Barroso5,6, Rubén García-Benito7, Anna Gallazzi11,12,Rosa M. González Delgado7, Simona Bekeraite8, Anna Pasquali10, Peter H. Johansson13, Begoña García Lorenzo5,6,

Glenn van de Ven14, Milena Pawlik1, Enrique Peréz7, Ana Monreal-Ibero8,15, Mariya Lyubenova14, Roberto CidFernandes16, Jairo Méndez-Abreu1, Jorge Barrera-Ballesteros5,6, Carolina Kehrig7, Jorge Iglesias-Páramo7,17,

Dominik J. Bomans18,19, Isabel Márquez7, Benjamin D. Johnson20, Robert C. Kennicutt21, Bernd Husemann9,8,Damian Mast23, Sebastian F. Sánchez7,17,22, C. Jakob Walcher8, João Alves24, Alfonso L. Aguerri5,6, Almudena

Alonso Herrero25, Joss Bland-Hawthorn26, Cristina Catalán-Torrecilla27, Estrella Florido28, Jean Michel Gomes29,Knud Jahnke14, Á.R. López-Sánchez32,33, Adriana de Lorenzo-Cáceres1, Raffaella A. Marino30, Esther

Mármol-Queraltó2, Patrick Olden1, Ascensión del Olmo7, Polychronis Papaderos29, Andreas Quirrenbach31, JoseM. Vílchez7, and Bodo Ziegler24

(Affiliations can be found after the references)

Received xx/xx/xxxx; accepted xx/xx/xxxx

ABSTRACT

We present optical integral field spectroscopy (IFS) observations ofthe Mice, a major merger between two massive (& 1011M⊙) gas-rich spirals NGC 4676A and B, observed between first passage and final coalescence. The spectra provide stellar and gas kinematics,ionised gas properties and stellar population diagnostics, over the full optical extent of both galaxies with∼1.6 kpc spatial resolution.The Mice galaxies provide a perfect case study highlighting the importanceof IFS data for improving our understanding of localgalaxies. The impact of first passage on the kinematics of the stars and gas has been significant, with strong bars likely induced inboth galaxies. The barred spiral NGC 4676B exhibits a strong twist in both itsstellar and ionised gas disk. The edge-on disk galaxyNGC 4676A appears to be bulge free, with a strong bar causing its “boxy”light profile. On the other hand, the impact of the mergeron the stellar populations has been minimal thus far. By combining the IFS data with archival multiwavelength observations we showthat star formation induced by the recent close passage has not contributed significantly to the global star formation rate or stellarmass of the galaxies. Both galaxies show bicones of high ionisation gas extending along their minor axes. In NGC 4676A the high gasvelocity dispersion and Seyfert-like line ratios at large scaleheight indicatea powerful outflow. Fast shocks (vs ∼350 km s−1) extendto ∼ 6.6 kpc above the disk plane. The measured ram pressure (P/k = 4.8× 106K cm−3) and mass outflow rate (∼ 8− 20M⊙yr−1) aresimilar to superwinds from local ultra-luminous infrared galaxies, although NGC 4676A has only a moderate infrared luminosity of3×1010L⊙. Energy beyond that provided by the mechanical energy of the starburst appears to be required to drive the outflow. Finally,we compare the observations to mock kinematic and stellar population maps extracted from a hydrodynamical merger simulation. Themodels show little enhancement in star formation during and following first passage, in agreement with the observations. We highlightareas where IFS data could help further constrain the models.

Key words. Galaxies: evolution, interactions, stellar content, ISM, Seyfert, kinematics and dynamics,nuclei,bulges; Techniques:Integral Field Spectroscopy

1. Introduction

NGC 4676A and B are members of the original Toomre (1977)sequence of merging galaxies, otherwise known as “the playingmice” (Vorontsov-Vel’Iaminov 1958). They are a classic exam-ple of a major gas-rich prograde merger, where the roughly equalmass of the progenitors and coincidence of the sense of rota-tion and orbital motion leads to lengthy tidal tails. NGC 4676Aand B (hereafter referred to as the Mice) are one of the earlieststage gas-rich major mergers visible in the nearby Universe, ob-served close to their apocentre, with N-body simulations and ob-servations agreeing that first passage must have occurred around170 Myr ago (Barnes 2004; Chien et al. 2007). The system isan outlying member of the Coma cluster, located about 4 or 1.7

⋆ Send correspondence to [email protected]

virial radii from the centre (Kubo et al. 2007), with a velocityabout 350 km s−1 from the mean cluster velocity (Burbidge &Burbidge 1961). Figure 1 shows an image of the Mice takenwith the Hubble Space Telescope (HST) Advanced Camera forSurveys (ACS).

Massive gas-rich galaxy mergers, of which the Mice are aclassic example, may play a key role in the evolution of thegalaxy population and in explaining the galaxy demographics inthe present day Universe. Within the currently favoured cosmo-logical model of a cold dark matter dominated Universe, struc-ture formation is hierarchical, with small overdensities formingearly on and subsequently merging to form larger structures.Galaxies form and evolve within these overdensities, or darkmatter halos. When the dark matter halos merge, the galaxies arethought to behave likewise, forming a single more massive sys-

Article number, page 1 of 23

Fig. 1. False colour image 2.4′×3.5′ in size in filters F606W (V,green channel) and F814W (I, red channel) for NGC 4676 using theHST-ACS/WFC (ACS Early Release Observations, Proposal ID 8992,P.I. Ford). Overlaid are V-band contours constructed from the CALIFAintegral field datacubes, with the outermost contour at 23 mag/arcsec2

and contours spaced by 0.45 mag/arcsec2. The inlaid schematic showsthe main kinematical properties of the Mice, as determined from long-slit spectroscopic observations (Burbidge & Burbidge 1961; Stockton1974), the CALIFA IFS data (Section 4), HI maps and comparisonswith N-body simulations (Barnes 2004). Black arrows indicate the di-rection of rotation of the disks, defined such that the disk of NGC 4676B(SE) is inclined away from the line-of-sight. NGC 4676A is viewed al-most exactly edge-on. The thick grey arrow indicates the approximatetrack that NGC 4676A (NW) has taken relative to NGC 4676B. Thepair are observed close to apocentre, with NGC 4676A receding fromNGC 4676B at∼160 km s−1 (Section 4).

tem. In order to obtain a complete understanding of how galax-ies formed and evolved within this gravitational framework, weneed to disentangle the relative importance of the many pro-cesses affecting the baryonic material.

By combining high quality spatially resolved optical spec-troscopy from the CALIFA survey with archival multiwave-length observations, the aim of this paper is to present a detailedpicture of the physical processes induced by the first passage oftwo massive gas-rich galaxies. In Figure 1 the contours of the re-constructed CALIFAV−band flux are superimposed on the HSTimage, showing how the large field-of-view of the integral-fieldunit used by the CALIFA survey allows coverage of the full ex-tent of the bodies of the Mice, including the inter-galaxy regionand some of the tidal tails. While the Mice are only one exampleof an early-stage gas-rich merger, and initial conditions of merg-ers (galaxy properties, impact parameter, orbital motionsetc.)

are expected to lead to a range of final outcomes, this study aimsto serve as a reference point for statistical studies of close galaxypairs, of particular use when combined with other similar casestudies (e.g. Engel et al. 2010 [NGC 6240]; Alonso-Herrero etal. 2012 [NGC 7771+7770]).

In the following section we collate published observed andderived properties of the Mice galaxies. Details of the observa-tions taken as part of the CALIFA survey are given in Section 3.We study the morphology, gas and stellar kinematics of the Micegalaxies in Section 4, including a new image decomposition ofthe archival HST images. In Section 5 we use the stellar contin-uum to constrain the star formation history of the galaxies.Wepresent maps of emission line strengths and ratios in Section 6.In Section 7 we collate multiwavelength observations from theliterature to obtain an accurate estimate of the ongoing star for-mation in the Mice galaxies. In Sections 8 and 9 we discuss theorigin of the high ionisation bicones in each of the galaxies. InSection 10 we present a mock integral field spectroscopy (IFS)datacube of the Mice merger, created from an hydrodynamicalsimulation, which we analyse using the same codes as the realdata cube. We collate the results in Section 11, to reveal theex-tent to which the interaction has affected the properties of theprogenitor galaxies.

Throughout the paper we assume a flat cosmology withΩM = 0.272,ΩΛ = 0.728 andH0 = 70.4km s−1Mpc−1 (WMAP-7)1. Masses assume a Salpeter initial mass function unless oth-erwise stated. From the redshift of the Mice we obtain a dis-tance of 95.5 Mpc, assuming negligible peculiar velocity contri-bution. The proximity of the Mice to the Coma cluster leadsto some uncertainty on their distance; unfortunately no redshift-independent distance measures are available. At a distanceof95.5 Mpc, 1′′ corresponds to 0.44 kpc and the effective spatialresolution of the CALIFA observations (3.7′′) is about 1.6 kpc.

2. Summary of previous observations of the Micepre-merger

As one member of the original Toomre (1977) merger sequence,the Mice have been observed at most wavelengths available toastronomers. The basic kinematics of the Mice were first stud-ied using long-slit spectroscopy covering the Hα+[N ii] lines(Burbidge & Burbidge 1961), who found the northern sides ofboth hulks to be receding. Stockton (1974) found Hα emit-ting gas in the northern tail to be receding from the body ofNGC 4676A with a velocity of 250 km s−1 at a distance of 90′′

(40 kpc). Barnes (2004) presented a hydrodynamic simulationof the Mice, where a reasonable match to the observed hydrogengas dynamics was obtained. In the remainder of this Section wegive a brief summary of the known physical properties of the twogalaxies, based on previous observations. These properties arecollated in Table 1. We focus on each galaxy in turn.

The northern NGC 4676A is a massive, gas-rich, activelystar-forming disk galaxy, viewed almost edge-on. It was classi-fied as an S0 galaxy by de Vaucouleurs et al. (1991), althoughwe suggest a revised classification below. From neutral Hydro-gen and optical observations, it is estimated to have a dynami-cal mass within the optical disk of 7.4×1010M⊙ (Hibbard & vanGorkom 1996). Molecular gas is primarily located in a centraldisk with scalelength of 2 kpc and thickness of 270 pc and thisdisk has a large molecular-to-dynamical mass ratio of 20% (Yun& Hibbard 2001). The two galaxies were not individually re-solved by IRAS, but recent Spitzer Space Telescope observations

1 http://lambda.gsfc.nasa.gov/product/map/current/best_params.cfm


Vivienne Wild et al.: The Mice at play in the CALIFA survey

Table 1.Properties of the Mice galaxies culled from the literature.

Parameter NGC 4676A (NW)/ IC 819 NGC 4676B (SE)/ IC 820RA (EquJ2000) 12h46m10.110s 12h46m11.237sDec (EquJ2000) +30d43m54.9s +30d43m21.87s

CALIFA ID 577 939SDSS objid 587739721900163101 587739721900163099Redshifta 0.02206 0.02204

r-band magnitudeb 13.22 13.03M(H i) [109M⊙]c 3.6 4.0M(H2) [109M⊙]d 5.7 3.6Mdyn (109M⊙e 74 129LFIR [1010L⊙]f 3.3 0.9

LX,0.5−2keV [1040erg s−1]g 0.6 1.2LX,2−10keV [1040erg s−1]g 0.9 1.5

Notes.Physical parameters have been converted to the same distance and Hubble parameter used in this paper (95.5 Mpc;H0 = 70.4km s−1Mpc−1).The unresolved pair is also known as Arp 242 and IRAS 12437+3059. (a) From H i 21cm (de Vaucouleurs et al. 1991).(b) From a growth curveanalysis of the SDSSr-band image (Walcher et al. in prep.). The SDSS catalogue magnitude for NGC 4676B is incorrect (see also Section 5.3).(c) Hibbard & van Gorkom (1996).(d) Yun & Hibbard (2001) for NGC 4676A and Casoli et al. (1991) for NGC4676B.(e) From optical extent andH i line width (Hibbard & van Gorkom 1996).(f) LFIR as defined by Helou et al. (1988). The Mice are not individually resolved by IRAS; theseestimates from Yun & Hibbard (2001) are from the total FIR flux and 1.4GHz radio continuum ratio.(g) González-Martín et al. (2009).

find that NGC 4676A accounts for 83% of the total 24µm flux ofthe system (Smith et al. 2007, and Section 7). Previous estimatesof its star formation rate (SFR) range from∼1 M⊙yr−1 from anarrow band Hα image where corrections for [Nii] emission,stellar absorption, dust attenuation or non-stellar emission werenot possible (Mihos et al. 1993), to 10 M⊙yr−1 using the unre-solved IRAS flux and resolved radio continuum (Hibbard & vanGorkom 1996; Yun & Hibbard 2001). We update this estimate inSection 7 using the full multi-wavelength information availableto us today. The steep mid-IR continuum of NGC 4676A sug-gests a high global heating intensity compared to normal star-forming galaxies, perhaps caused by a nuclear starburst (Daleet al. 2000; Haan et al. 2011).

NGC 4676A has LINER-like optical emission line ratios(Keel et al. 1985), but has a diffuse X-ray morphology, leadingGonzález-Martín et al. (2009) to conclude that NGC 4676A doesnot harbour an AGN. Plumes of Hα extending along the minor-axis of NGC 4676A were identified by Hibbard & van Gorkom(1996). These have been linked to outflowing gas from the nu-cleus, coincident with diffuse soft X-ray emission(Read 2003).The unusually high ionised-to-neutral (7.7/11.3µm) PAH ratiofound to the east of the nucleus by Haan et al. (2011) may alsobe associated with this outflow, or its driving source.

The south-east NGC 4676B is a massive, strongly barred spi-ral galaxy. Its faintness at FIR wavelengths indicates a loweroverall star formation rate than its partner, and the mid-IRcon-tinuum shape suggests a star formation intensity consistent withnormal star-forming galaxies (Dale et al. 2000; Haan et al. 2011).The dynamical mass within the optical disk is measured to be1.29×1011M⊙ (Hibbard & van Gorkom 1996), about 50% largerthan that of NGC 4676A. Concentrations of molecular gas arelocated at either end of the strong bar (Yun & Hibbard 2001).Single-dish CO observations by Casoli et al. (1991) result in aCO line flux that is a factor of 2.5 greater than the interferomet-ric measurement of Yun & Hibbard (2001), which suggests thata large fraction of the molecular gas in NGC 4676B is eitherlow surface brightness or extended over scales larger than thoseprobed by the interferometric observations (θ > 45′′, 20 kpc).The total Hydrogen mass of NGC 4676B is a little lower thanthat of NGC 4676A (Hibbard & van Gorkom 1996); given its

larger dynamical mass, NGC 4676B has a significantly lowergas mass fraction than NGC 4676A.

With a LINER-like emission line spectrum (Keel et al. 1985),compact X-ray emission and a detection in hard X-rays (L2−10 =

1.48× 1040erg s−1), NGC 4676B is classified as a “candidate”AGN by González-Martín et al. (2009). Additional evidence forthe presence of an AGN comes from: (i) the unusually high ra-tio of excited H2 emission to PAH emission (Haan et al. 2011),consistent with additional excitation of H2 by hard X-rays froma central AGN (Roussel et al. 2007); (ii) the hard X-ray emissionis offset from the nuclear Hα emission (Masegosa et al. 2011),showing that the hard X-rays are not related to a nuclear star-burst.

In summary, previous observations have shown thatNGC 4676A is an edge-on disk, rich in dust, molecular andatomic gas, while NGC 4676B is a strongly barred, inclined disk,with a lower gas mass fraction. There is some evidence for a lowluminosity AGN in NGC 4676B, but no evidence for an AGN inNGC 4676A. A bipolar outflow is found in NGC 4676A. Table 1summarises the properties of the Mice galaxies that can be foundin the literature.

3. The CALIFA Data

The Calar Alto Legacy Integral Field Area (CALIFA) survey,currently in progress on the 3.5m Calar Alto telescope in Spain,aims to obtain spatially resolved spectra of 600 local galax-ies spanning the full colour-magnitude diagram, with the PPAKintegral field unit (IFU) of the PMAS instrument (Roth et al.2005). The galaxies are diameter selected to be volume cor-rectable from the seventh data release (DR7) of the Sloan DigitalSky Survey (SDSS). Observations are performed using two over-lapping grating setups (V500 and V1200), with resolutions of6.3Å and 2.3Å (FWHM) and wavelength ranges of 3745–7500Åand 3650–4840Å respectively (velocity resolution,σ ∼100–210 km s−1 and 60–80 km s−1). The spectrophotometric calibra-tion accuracy is close to that achieved by the SDSS DR7. Thelarge field-of-view of PPAK (1.3′) allows coverage of the fulloptical extent of the galaxies. The datacubes reach a 3σ limit-ing surface brightness of 23.0 mag/arcsec2 for the V500 grating


Fig. 2. The Hα map of the Mice galaxies with examples of V500 CALIFA spectra, extractedfrom key regions. For the nucleus and knots ofboth galaxies the spectra are extracted using a circular aperture of 2′′ radius, for the tail and outflow regions in NGC 4676A an elliptical apertureof 4′′×3′′ was extracted.

data and 22.8 mag/arcsec2 for V1200. The survey is described indetail in the presentation article (Sánchez et al. 2012a) and firstData Release article (DR1, Husemann et al. 2013).

Observations of the Mice galaxies were taken during thenights of 2011-05-04 (V500), 2011-05-05 (V1200) and 2011-06-29 (V1200). Three dithered pointings were made, centredaround each of the two galaxy nuclei, with exposure times of900 s per pointing for V500 and 2× 900 s for V1200. The PPAKIFU data of the Mice were reduced with the dedicated CALIFAdata reduction pipeline and are available to the public as partof DR1. The basic outline of the pipeline is given in Sánchezet al. (2012a), with significant improvements made for the DR1release and described in Husemann et al. (2013). In additiontothe data available in the public release, a combined cube withobservations from both gratings was created to improve the vi-gnetting of the blue and red end of the V500 and V1200 datarespectively. Wavelengths shorter (longer) than 4500Å originatefrom the V1200 (V500) data, with the V1200 data cube degradedto the spectral resolution of the V500 data. The V500 and de-graded V1200 cubes were then spatially co-registered during theprocess of differential atmospheric refraction (DAR) correctionand their relative spectrophotometry matched in the overlappingspectral regions before being combined into a single cube.

Finally, for the purposes of presentation in this paper, thecubes for NGC 4676A and NGC 4676B were cropped diago-nally between the galaxies and joined to form a single cube, withthe relative offset between the two nuclei determined from theirSDSS centroids. This procedure is only astrometrically accurateto within the size of the spaxels, i.e. 1′′, which was deemed suf-ficient given the spatial resolution of the data of∼3.7′′. Voronoibinning was performed on the individual cubes, in order to retainindependence of the two sets of observations.

Figure 2 presents the Hα emission line intensity map of theMice galaxies, measured from the V500 data, and provides ex-amples of spectra extracted from key regions of interest.

4. Morphology and kinematics

We begin by combining archival HST imaging and CALIFAspectroscopy to build a global picture of the morphology andkinematics of the merging galaxies.

4.1. Image decomposition

To measure the main morphological components of the Micegalaxies we use GALFIT 3.0 (Peng et al. 2010) on the archivalHST ACS F814W image. The instrumental point-spread-function (PSF) was synthesised using TinyTim; Galactic stars,tidal arms and prominent dust lanes in NGC 4676A were maskedusing SExtractor. The parameters of the fit are presented in Table2.

We find that NGC 4676A is dominated by an edge-on diskand/or bar. No evidence could be found for either a nuclear pointsource or bulge, although the central dust lanes may still hidethese components. The boxy shape apparent in the HST imagecould arise from vertical motions of stars in a bar, rather than abulge (e.g. Bureau & Athanassoula 1999, 2005; Kuijken & Mer-rifield 1995; Lang et al. 2014; Martínez-Valpuesta et al. 2006;Williams et al. 2011), and the model fit improves upon inclusionof a Ferrer bar component. Our results show that the previousclassification of NGC 4676A as an S0 galaxy is likely incorrect(de Vaucouleurs et al. 1991), and SBd is more appropriate. Thisrevised classification is also consistent with the high dustand gascontent of the galaxy.

The inclination of NGC 4676B allows the identification of astrong bar, although our simple image decomposition is unableto separate the bar from the disk component. By ellipse fittingthe HST image isophotes using the method presented in Aguerriet al. (2009), we find the radius of the bar in NGC 4676B tobe 10′′ (4.4 kpc) and position angle (PA)∼20. This is consis-tent with the peaks of CO, separated by∼ 16′′ and thought tolie at either end of the bar. A significant bulge is required toobtain a good fit. Including a nuclear point-source componentalso improves the fit slightly and decreases the Sérsic indexofthe bulge as expected (Gadotti 2008). The best fit model has abulge-to-total flux ratio of∼ 0.5 i.e. an S0/a galaxy (Simien &de Vaucouleurs 1986).

4.2. Stellar kinematics

The stellar kinematics of the Mice galaxies are measured fromthe V1200 grating datacube using the penalised pixel-fitting(pPXF) method of Cappellari & Emsellem (2004). Full detailsof the method will be given in Falcón-Barroso et al. (in prep.).



Table 2. Results of a GALFIT 3.0 decomposition of the HST ACS F814W image.

Component magnitude parametersA : Edge-on Disk 14.7:a edgedisk:µ0=21.8 ; hs=3.5 ;rs=29: ; PA= 5A : Bar 14.1a ferrer2:µFWHM=19.7 ; Rtrunc=19;α=2; β=1; b/a=0.6; PA= 4B : Bulge 14.2 Sérsic: Re=6.0: ; n=4.3: ; b/a=0.75 ; PA=-26B : Disk+Bar 13.8 expdisk: Rs=6.2 ; b/a=0.4 ; PA= 36B : Point source 20.9: PSF

Notes.Parameters are given as described in the GALFIT documentation. Length scales are in arcseconds, position angles are from N to E (anti-clockwise in the figures) and surface brightnesses are in mag/arcsec2. Parameter values should be taken as indicative only, and those markedwith“:” are particularly uncertain due to the complex morphology of the system.(a) Measured from the model image.

Briefly, a non-negative linear combination of a subset of 328stellar templates from the Indo-US library (Valdes et al. 2004)covering a full range of stellar parameters (Te f f , log(g), [Fe/H])is fit to the spectra, following a Voronoi binning of spaxels toachieve a minimum signal-to-noise (SNR) of 20 (Cappellari &Copin 2003). These Voronoi binned spaxels are termed “vox-els”, and the effective spatial resolution of Voronoi binned mapsis naturally degraded below the CALIFA effective resolution of3.7′′in the outer regions. In the computation of the SNR we havetaken into account the correlation among the different spectra inthe datacube as explained in Husemann et al. (2013). Spaxelsinthe original V1200 datacube with per pixel SNR< 3 are deemedunreliable and not included in the analysis. Emission linesin thecovered wavelength range are masked during the fitting proce-dure (i.e. [Oii], [Ne iii], Hδ, Hγ, [O iii], He ii, and [Ar iv]). Errorestimates are determined via Monte Carlo simulations, and aretypically 5 (10)km s−1 for velocities and dispersions in the in-ner (outer) voxels. Velocity dispersions are corrected forinstru-mental resolution during the fitting process, velocity dispersionsbelow∼35km s−1 are unresolved at the resolution of the data.

The stellar velocity and velocity dispersion maps are shownin the top panels of Figure 3, where the systemic redshift ofthe system is taken as the velocity of the stars in the central5′′ of NGC 4676A. The nucleus of NGC 4676A is receding at∼160 km s−1 relative to the nucleus of NGC 4676B. The stellarkinematics of the main body of NGC 4676A show a rotatingedge-on disk, with rotation axis coincident with the minor diskaxis. This rotation continues into the northern tidal tail,whichis receding at 130 km s−1 relative to the nucleus of NGC 4676Ain the northernmost voxel (30-35′′ from the nucleus). No evi-dence for a classical bulge is seen in this galaxy, with low nu-clear velocity dispersion and constant rotation with height abovethe major axis. This supports the results from the image analysisabove. However, given the disturbed nature of the system andthe high dust attenuation close to the nucleus, high spatialreso-lution longer wavelength observations would be required togivea robust upper limit on bulge size. We note that we are unable toidentify “pseudo” bulges at the spatial resolution of the CALIFAdata.

The stellar velocity field of NGC 4676B shows a twisted in-clined disk (Z or S-shaped isovelocity contours), with the in-ner rotation axis offset from the minor axis of the disk. A di-rect analysis of the kinematic maps provides a quantitativemea-surement of the kinematic centres and PAs of the galaxies (seeBarrera-Ballesteros et al. 2014, for details of the method). Fora radius internal to 10′′ we measure receding and approachingstellar kinematic PAs of 165 and 150 degrees respectively (anti-clockwise from north), compared to a morphological PA of 33degrees, confirming that the dominant rotation in the inner re-gions is around the major axis of the galaxy. The classical bulgein NGC 4676B is clearly visible from the significant increasein

velocity dispersion in the centre of the galaxy, confirming theresults from the morphological decomposition.

4.3. Ionised gas kinematics

We measure the ionised gas kinematics from the Hα line in theV500 datacube, favouring the increased SNR provided by thestronger Hα line, over the higher velocity resolution afforded bythe weaker lines in the V1200 cube. We verified that our re-sults are consistent with those derived from [Oii] in the higherspectral resolution V1200 datacube in the inner high surfacebrightness regions. We use standard IRAF routines to fit si-multaneously three Gaussian line profiles to Hα and the two[N ii] lines, with both amplitude and width free to vary indepen-dently, in every spaxel that is included in the Hα emission linequality mask described in Section 6. The instrumental resolu-tion was subtracted from the measured line widths in quadrature(σ =116 km s−1 at the wavelength of Hα). Typical errors on theline velocities were estimated to be∼15 km s−1 from fitting dif-ferent species. Line widths below∼58km s−1 are unresolved atthe CALIFA resolution.

The ionised gas kinematics derived from the Hα emissionline are shown in the bottom panels of Figure 3. The same globalkinematics are seen as in the stellar kinematics: NGC 4676Ais dominated by a rotating edge-on disk and NGC 4676B by atwisted disk inclined to the line-of-sight. Evidence of a dynam-ically hot bulge is seen in NGC 4676B, but not in NGC 4676A.The receding and approaching kinematic PAs of the ionised gasdisk in NGC 4676B are 350 and 161 degrees respectively, con-sistent with those of the stars i.e. the twist in the disk is asstrongin both the stellar and ionised gas kinematic fields. The ionisedgas in the N-E spiral arm of NGC 4676B and northern tidal tail ofNGC 4676A are the dynamically coldest regions of the systemwith observed line widths close to the resolution of the CAL-IFA data, implying velocity dispersions below∼58km s−1. Theionised gas in the tail of NGC 4676A is receding at∼180 km s−1,relative to the body of NGC 4676A, at a distance of∼35′′. Thisis in agreement with long-slit observations by Stockton (1974).

The zero-velocity curve of the ionised gas in NGC 4676Ashows a V-shape along the minor axis, indicating lower line-of-sight velocities above the plane of the disk than within the disk.This extra-planar gas also shows enhanced velocity dispersion,in streamers extending radially outwards from the nucleus in thesame direction as the soft X-ray emission (Read 2003). Theseare dynamical signatures of outflowing gas, i.e. a galactic super-wind, which we will return to discuss in more detail in Section 8.Additional regions of high velocity dispersion are observed closeto the nucleus of NGC 4676B. In this case, the velocity gradientis high and this effect might be equivalent to “beam smearing”


20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

∆ v (km/s)

-300

-200

-100

0

100

200

N

E

Stars

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

σv (km/s)

0

55

110

165

220

N

E

Stars

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

∆ v (km/s)

-300

-200

-100

0

100

200

N

E

Gas

20 0 -20∆α (arcsec)

-20

0

20

40

60

80∆δ

(ar

csec

)σv (km/s)

0

55

110

165

220

N

E

Gas

Fig. 3. Stellar (top) and ionised gas (bottom) velocity field (left) and velocity dispersion (right). The ionised gas maps are measured from the Hα

emission line in the V500 grating. Velocities are relative to 6652 km s−1 which corresponds to the heliocentric velocity of stars measured within thecentral 5′′ of NGC 4676A. The velocity dispersion maps have been corrected for instrumental resolution; velocity dispersions below∼35km s−1

for stars and∼58km s−1 for the gas are unresolved at the instrumental resolution. Typical errors are 5-10km s−1 for the stellar velocities and15km s−1 for the ionised gas. In these maps and all the following maps, the black contours indicateV-band isophotes with the outermost contourat 23 mag/arcsec2 and contours spaced by 0.6 mag/arcsec2. The black circle shows the effective spatial resolution of the CALIFA observationsbefore Voronoi binning of the data.

seen in Hi surveys, where multiple components are observed ina single resolution element (3.7′′ in the case of CALIFA).

We detect a blueshift of 30km s−1 of the gas relative to starsin the nucleus of NGC 4676A, and an 88km s−1 redshift of thegas relative to the stars in the nucleus of NGC 4676B. In thecase of NGC 4676A this is a clear case for an outflow withsome transverse velocity component. The cause of the offset forNGC 4676B is less obvious, but may be due to a tidal compo-nent passing in front of the bulge and contributing significantlyto the Hα luminosity.

The spectral resolution of CALIFA is not sufficient to iden-tify dynamically distinct components in the emission lines, how-ever measurements of line asymmetries indicate where suchmultiple components may exist. Following the cross-correlationmethod of García-Lorenzo (2013) for the [Oiii] emission line we

find a mixture of blue- and redshifted asymmetries in the west-ern bicone of NGC 4676A, again consistent with gas outflowingin bicones perpendicular to the disk.

5. Stellar populations

In this section we use the stellar continuum shape and strengthof stellar absorption features to characterise the spatially re-solved properties of the stellar population and constrain the stel-lar masses and star formation history of each galaxy. From thesimulations of Barnes (2004) we expect first passage to have oc-curred about 170 Myr ago, and optical spectra of star clustersindicate some star formation occurred at that time (Chien etal.2007). Here we investigate how wide spread that star formationwas and whether it has continued to the time of observation.



20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

D4000n

1.00

1.20

1.40

1.60

1.80

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

W(HδA+HγA)

-5.48

-1.38

2.72

6.82

10.91

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

[Mg 2Fe]

0.30

0.35

0.40

0.45

0.50

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log(t Lr/yr)

8.90

9.15

9.40

9.65

9.90

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log(t M*/yr)

8.90

9.15

9.40

9.65

9.90

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log(Z/Z Ο •)

-0.77

-0.48

-0.19

0.10

0.39

N

E

Fig. 4. Top: Maps of three commonly used spectral line indices: the 4000Å break strength (Dn4000); the combined equivalent width ofthe higher-order Balmer absorption lines HδA and HγA(in Å); the Lick index [Mg2Fe]. Bottom: Mean light-weighted age in ther-band, stellarmass-weighted age, and stellar metallicity. These have been measured from comparison of the line indices to stellar population synthesis models.

The stellar population analysis is performed on the combinedV500 and V1200 cube, using the same Voronoi binning as for thestellar kinematics. While it is difficult to quantify the uncertain-ties associated with the extraction of physical propertiesfromstellar population model fitting (see e.g. Conroy et al. 2009; Pan-ter et al. 2007), Cid Fernandes et al. (2014) present an analysisof the errors caused by use of different spectral synthesis mod-els and spectrophotometric calibration, applied to the CALIFAdata. They find that noise and shape-related errors at the levelexpected for CALIFA lead to uncertainties of 0.10–0.15 dex instellar masses, light-weighted mean ages and metallicities, withlarger uncertainties on star formation histories and therefore onmass-weighted quantities. There are even larger uncertaintiesassociated with the choice of population synthesis model, at thelevel of 0.2-0.3 dex. For this reason, we present raw line indexmaps alongside derived quantities.

5.1. Spectral indices

We measure five standard stellar absorption line indices fromemission line subtracted spectra: the 4000Å-break (D4000n), theBalmer absorption lines Hβ and HδA+HγA , and two compositemetallicity-sensitive indices [Mg2Fe] and [MgFe]’ as defined inBruzual & Charlot (2003) and Thomas et al. (2003), respectively.Emission lines detected at greater than 3σ significance are sub-

tracted from the stellar continuum using a Gaussian broadenedtemplate. Physical properties are then derived through compar-ison of the line indices to a “stochastic burst” library of stellarpopulation synthesis models (Bruzual & Charlot 2003), follow-ing the same Bayesian approach as described in Gallazzi et al.(2005). Typical errors on mean ages are∼0.12dex, varying be-tween 0.1 and 0.2 dex.

Maps of three of the measured spectral indices are shownin the upper panels of Figure 4. Derived properties of light-weighted and mass-weighted mean stellar age and metallicityare shown in the bottom panels. The strength of the 4000Å break(top left) correlates to first-order with mean stellar age, averagedover timescales of several Gyr, and to second-order with metal-licity, particularly in the older stellar populations. Thestrengthof the higher order Balmer absorption lines (top centre) measuresmean stellar age over a slightly shorter period of∼ 0.5 Gyr. Boththe central region of NGC 4676B (in the bulge), together withthe east and west flanks of NGC 4676A (above and below thedisk), show strong break strengths indicating older mean stel-lar ages than in the disks of the two galaxies. The region withthe youngest mean stellar age is the north-east (NE) tidal arm ofNGC 4676B. In star-forming galaxies the strength of the Balmerlines and 4000Å break strength are strongly inversely correlated.It is therefore not surprising that (the inverse of) the Balmer lineindex map is not dissimilar to that of D4000, although with a


20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

fL,young

0.0

0.2

0.4

0.6

0.8

1.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

fL,intermediate

0.0

0.2

0.4

0.6

0.8

1.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

fL,old

0.0

0.2

0.4

0.6

0.8

1.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

fM*,young

0.0

0.2

0.4

0.6

0.8

1.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

fM*,intermediate

0.0

0.2

0.4

0.6

0.8

1.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

fM*,old

0.0

0.2

0.4

0.6

0.8

1.0

N

E

Fig. 5. The fraction of light at 5635Å (top panels) and mass (lower panels) contributed by young (t <140 Myr), intermediate (140 Myr<t <1.4 Gyr) and old (t >1.4 Gyr) populations.

slightly lower SNR2. The right hand panel shows the metallicitysensitive index, [Mg2Fe]. Both the bulge of NGC 4676B andregions above and below the disk plane of NGC 4676A showstronger [Mg2Fe] than the disks, indicating a more metal richstellar population. The index [MgFe]’ shows similar results.

Converting these observables into physical properties in thelower panels of Figure 4, we find that the stars of both galaxiesare predominantly older than several Gyrs. Only the northerntidal tail and the NE arm of NGC 4676B have distinctly youngerstellar populations, with light-weighted ages of∼ 0.6 Gyr ormass-weighted ages of∼ 2.5 Gyr. The bulge of NGC 4676Band regions above and below the disk of NGC 4676A have theoldest mean stellar ages of& 4.5 Gyr. There is no evidence fora significant intermediate age population that would be consis-tent with having formed during first passage, therefore it isclearthat the merger has not yet had a significant impact on the stellarpopulations.

2 Note that the negative equivalent widths of the Balmer lines do notindicate emission, measured values as low as−10 are expected for oldstellar populations (see Gallazzi et al. 2005, for the expected values forSDSS galaxies).

5.2. Young, intermediate age and old populations

To visualise the spatially resolved star formation historyof thegalaxies, we turn to the full-spectrum fitting packagestarlight(Cid Fernandes et al. 2005, 2013). This inverts the observedspectrum into stellar populations of different ages, rather thanfitting a library of models with analytic star formation histo-ries, as was done in the previous section.starlight fits thefull wavelength range with combinations of simple stellar pop-ulation (SSP) spectra from the population synthesis modelsofVazdekis et al. (2010) and González Delgado et al. (2005), us-ing the Granada (Martins et al. 2005) and MILES (Sánchez-Blázquez et al. 2006) stellar libraries, dust extinction followingthe Cardelli et al. (1989) law, a Salpeter IMF, and stellar evolu-tionary tracks from Girardi et al. (2000). The SSP ages rangefrom 1 Myr to 14 Gyr and SSPs of four different metallicities areincluded (Z =0.0004, 0.008, 0.020, 0.033, whereZ⊙ ∼ 0.02 forthese models). Emission lines are masked during the fit, anderrors propagated from the CALIFA error arrays.

Figure 5 shows maps of the fraction of light at 5635Å andmass arising from stars of different ages. We select age binsthat correspond to the main sequence lifetimes of stars withdistinctly different optical line and continuum features, namelyyoung (t <140 Myr), intermediate (140 Myr< t <1.4 Gyr) andold (t >1.4 Gyr) populations, in order to maximise the robust-ness of the spectral decomposition results.



20 0 -20∆α (arcsec)

-20

0

20

40

60

80∆δ

(ar

csec

)

log[f(H α)]

-0.5

0.0

0.5

1.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

AV,gas

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log[I(H α)]

3.0

4.0

5.0

6.0

7.0

N

E

Fig. 6. Left: Hα emission line flux (10−16erg s−1 cm−2 arcsec−2). Centre: Emission line attenuation at 5500Å (magnitudes), calculated from theobserved Hα/Hβ emission line ratio and dust attenuation law suitable for emission lines in low redshift star-forming galaxies (Wild et al. 2011).Right: dust attenuation corrected Hα luminosity surface density (L⊙ kpc−2).

Clearly the ongoing merger has thus far had little significantaffect on the stellar populations in terms of total stellar massorglobal star formation history. The fraction of stellar masscon-tributed by stars younger than 140 Myr is less that 5% in allregions, although∼30% of optical light in the nuclear regionsof NGC 4676A and tidal arm of NGC 4676B arises from starsyounger than 140 Myr. In most regions> 90% of the stellar massis from stars older than 1.4 Gyr. There is an excess of light fromintermediate age stars seen in the western half of NGC 4676A,inter-galaxy region, and NE bar of NGC 4676B, compared tothe disks of these galaxies. This is consistent with the triggeringof low level star formation in the gas flung out from the galaxydisks at first passage∼170 Myr ago, perhaps through dynamicinstabilities or shocks (see e.g. Boquien et al. 2010, and refer-ences therein).

5.3. Stellar mass and star formation rate

From the integrated star formation history measured bystarlight, and correcting for recycling of matter back into the in-terstellar medium (ISM), we obtain a total current stellar mass of1.2×1011 and 1.5×1011M⊙ for NGC 4676A and B respectively,for a Salpeter IMF. For NGC 4676A the CALIFA derived stellarmass agrees well with that derived from SDSS 5-band photome-try (1.6× 1011M⊙, J. Brinchmann3). For NGC 4676B the SDSSderived mass is 10 times lower, which is traceable to incorrectphotometric measurements in all bands in the SDSS catalogue,presumably caused by the deblending algorithm which has iden-tified three sources in this galaxy. The higher stellar masses thandynamical masses (Table 1) may be due to the assumed SalpeterIMF or difficulty in measuring a dynamical mass from a kine-matically disturbed system. The use of a Chabrier IMF wouldsolve the discrepancy.

3 The SDSS catalogue can be downloaded from here:http://home.strw.leidenuniv.nl/∼jarle/SDSS/DR7/totlgm_dr7_v5_2.fit.gzSDSS mass was increased by a factor of 1.8 to convert from a Chabrierto Salpeter IMF.

The decomposition of the stellar continuum can provide anapproximate estimate of the ongoing star formation rate in thegalaxies, independent of the ionised gas emission. The valueobtained depends sensitively on the width of the time bin overwhich the average is calculated. Varying the width of the timebin between 25 and 140 Myr yields a global SFR of 2.6-5 M⊙yr−1

and 1-4 M⊙yr−1 for NGC 4676A and B respectively. While theerrors on these values are difficult to quantify, it is clear thatneither galaxy is currently undergoing a significant burst of star-formation, especially when their very large stellar massesaretaken into account.

The young stellar population found in the centre ofNGC 4676A implies a nuclear SFR surface density of∼0.15M⊙ yr−1 kpc−2, averaged over 140 Myr in the inner 5× 5′′.No young stellar population is found in the central regions ofNGC 4676B.

6. Ionised gas emission

One of the highlights of the CALIFA dataset is its wide wave-length coverage, allowing measurements of all the strong emis-sion line species from [Oii]λ3727 to [Sii]λ6731. We measurethe total emission line fluxes in each spaxel by fitting Gaussianline profiles to the stellar continuum subtracted spectra using theIFU package fit3D (see e.g. Sánchez et al. 2007).

We use empirically derived flux thresholds to remove spax-els from the analysis where the surface brightness is too lowtoobtain a reliable line flux. For Hα and line ratios which includeHα we use a flux threshold of 1.5 × 10−17erg s−1 cm−2, and forHβ and line ratios which include Hβ we use a flux threshold of5 × 10−18erg s−1 cm−2. Low SNR spectra at the outskirts of themaps are visually checked to ensure that their line measurementsare reasonable. In order to measure emission line strengthsinthe fainter outer regions, we performed the same analysis onaVoronoi-binned data cube. However, we found that no signif-icant additional information was gained from these maps, andsome information was lost due to the lack of connection betweenVoronoi bin boundaries and physical components.


20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log[f([NII])/f(H α)]

-0.4

-0.3

-0.2

-0.1

0.0

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log[f([SII])/f(H α)]

-0.6

-0.5

-0.4

-0.3

-0.2

-0.1

-0.0

0.1

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log[f([OIII])/f(H β)]

-0.4

-0.2

-0.0

0.2

0.4

0.6

N

E

20 0 -20∆α (arcsec)

-20

0

20

40

60

80∆δ

(ar

csec

)log[f([OI])/f(H α)]

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

N

E

Fig. 7. From top left to bottom right: emission line flux ratio maps for [Nii]λ6583/Hα, ([S ii]λ6717+[S ii]λ6731)/Hα, [O iii]λ5007/Hβ and[O i]λ6300/Hα. To guide the eye in comparison between panels the black contours indicateV-band isophotes, and the thick green contoursdelineate those regions with log([Nii]/Hα)> −0.25.

6.1. Hα emission and dust attenuation

The CALIFA datacubes have sufficient continuum SNR, spec-tral resolution and wavelength range to allow subtraction of thestellar absorption from the Balmer emission lines. This allowsus to measure the dust attenuation affecting the nebular linesand thus estimate the intrinsic distribution of ionised gas. Thedust attenuation (AV,gas) map is constructed from the Hα/Hβ ra-tio map using the attenuation law from Wild et al. (2011) whichis measured from, and applicable to, emission lines in localstar-forming galaxies including local ULIRGs.

Figure 6 shows the observed Hα line flux, the emission lineattenuation map (AV,gas) and the dust-attenuation corrected (in-trinsic) Hα line luminosity. The three peaks in observed Hα fluxalong the disk of NGC 4676A have been noted previously in theliterature. The ionised gas bar of NGC 4676B is clear, with theaxis of the bar offset by about 25 in the clockwise direction fromthe major axis of the continuum light profile (contours).

We measure a maximum effective attenuation of∼ 7 mag-nitudes at 5500Å close to the centre of NGC 4676A, which isconsistent with the dust lane visible in the HST images. Forsuch large attenuations, small variations in the assumed dust at-tenuation law or stellar population models result in significantuncertainties in the dust-corrected line fluxes. We show belowthat the dust attenuation corrected total Hα flux is a factor ofa few larger than expected from multiwavelength observations(Section 7 and Table 3). After correcting for dust attenuation wefind that ionised hydrogen emission in NGC 4676A is concen-trated in the central regions of the galaxy.

NGC 4676B has an average line extinction of a little over∼1 magnitude, typical for an ordinary star-forming galaxy. Thedust content of the disk appears slightly higher than the bar. Wenote that Hα emission is seen even in the region dominated bythe old stellar bulge. Either there are sufficient young stars areavailable to ionise the gas, even though the continuum lightiscompletely dominated by old stars (Section 5), or the gas in the



20 0 -20∆α (arcsec)

-20

0

20

40

60

80∆δ

(ar

csec

)f([SII] 6717)/f([SII] 6731)

0.7

0.8

0.9

1.0

1.1

1.2

1.3

1.4

N

E

Fig. 8. Map of the [Sii] doublet ratio ([Sii]λ6717/[S ii]λ6731), asensitive electron density estimator. Higher density gas has lower ratios.The thin black contours show V-band continuum flux, and the thickgreen contours delineate those regions with log([Nii]/Hα)> −0.25.

central regions is primarily ionised by the AGN. Alternatively,the ionised gas is not coincident with the bulge, which wouldalso be consistent with the offset in velocities (Section 4)

6.2. Emission line ratios sensitive to ionisation source

Standard line ratios sensitive to the shape of the ionising spec-trum have been calculated and the spatial distribution of a se-lection of these are shown in Figure 7. We focus on pairs oflines that are close enough in wavelength space for their ratiosto not be strongly affected by dust attenuation. The most ob-vious feature of these maps are the butterfly shaped bicones ofhigher ionisation gas orientated along the minor axes of bothgalaxies. Comparing the four line ratio maps, the bicones inthetwo galaxies are clearly different: in NGC 4676A the biconesare visible in all maps, in NGC 4676B the bicones have higher[S ii]/Hα, [N ii]/Hα, and [Oi]/Hα line ratios, but are not visiblein [O iii]/Hβ. High ionisation gas is also found following an arcto the south of NGC 4676A.

The line ratios make it clear that a substantial fraction of theline emission originates from sources other than photo-ionisationby stars. To guide the eye, we overplot contours to delineatethose regions with log([Nii]λ6584/Hα)> −0.25, which is ap-proximately the maximum ratio observed in high-metallicity starforming regions (Kauffmann et al. 2003). We will analyse thepossible causes of these bicones in Sections 8 and 9 below.

6.3. Emission line ratios sensitive to gas density

In Figure 8 we show the [Sii]λ6717/[S ii]λ6731 doublet ra-tio which is primarily sensitive to the electron density of theemitting gas (ne). This ratio varies between 0.4 and 1.4 forne & 105cm−3 and ne . 100cm−3 (at Te = 104K, Osterbrock& Ferland 2006). Both galaxies show complex structure in their

gas density. Regions with particularly low density (high doubletratio) are the NE tidal arm of NGC 4676B, and the edges of thewestern bicone in NGC 4676A. The apparently very high den-sity in the outskirts of NGC 4676B and the region to the southof NGC 4676A might be due to shocks caused by the interactionand gas inflows, but higher quality data should be obtained toconfirm these measurements.

The median doublet ratio in the high ionisation bicones ofNGC 4676A is 1.1, which implies a typical electron density ofne ∼ 400cm−3. In the very outer regions of the western cone,the median [Sii] ratio increases slightly to 1.2, equivalent to anelectron density ofne ∼ 200cm−3. The central electron densityis no higher than in the gas throughout the cones.

6.4. Gas-phase metallicities and star formation rates

We calculated metallicities from the [Oiii] and [N ii] emissionlines (Pettini & Pagel 2004) in regions where the line ratiosin-dicate the gas is primarily photoionised by hot stars, i.e. in thedisks of the galaxies and in the northern tidal tail. The metal-licity of the gas throughout both galaxies is approximatelysolar,consistent with star clusters located in the tidal tails of the twogalaxies (Chien et al. 2007).

We additionally note that the [Nii]/Hα ratio map is notice-ably flat in the disk and bar of NGC 4676B. Given that thisgalaxy is massive, with a prominent central bulge, a metallicitygradient is expected (Pilyugin et al. 2004; Sánchez et al. 2012b).Flat gradients can be caused by mergers, but such a flat gradientis expected to only occur in the later stages of a merger (Kewleyet al. 2010; Rupke et al. 2010a,b). The presence of a strong barcomplicates the picture, and simulations by Martel et al. (2013)show that mixing of gas from the outer regions to the centrethrough a bar may occur even before star formation is induced.

For the regions of the galaxies where the gas is photoionisedby light from stars, we can measure the SFR from the dust at-tenuation corrected Hα line luminosity, using the standard con-version to star formation rate (Kennicutt 1998b). For the re-maining regions we obtain only an upper limit on the SFR, be-cause a fraction of the Hα flux must arise from processes un-related to star formation. For NGC 4676A we measure a totalHα luminosity of 4.4 × 1042erg s−1, of which 2.2 × 1042erg s−1

(50%) arises from regions with line ratios consistent with stellarphotoionisation. These lower and upper limits lead to a SFRfor NGC 4676A of 17-35 M⊙yr−1. In the same manner, forNGC 4676B we measure a total dust-attenuation corrected Hαluminosity of 6.1 × 1041erg s−1 of which 56% arises from re-gions with line ratios consistent with photoionisation by youngstars. This gives a SFR for NGC 4676B of 3-5 M⊙yr−1.

We can use the dust attenuation corrected Hα luminosity toestimate the central SFR surface density in the inner 5× 5′′ ofeach galaxy. For NGC 4676A we findΣSFR < 4M⊙ yr−1 kpc−2,and for NGC 4676B we findΣSFR < 0.2M⊙ yr−1 kpc−2. Theupper limit estimated from the dust corrected Hα luminos-ity is similar to that measured from the spectral decomposi-tion for NGC 4676B (∼ 0M⊙ yr−1 kpc−2), but much higher forNGC 4676A (∼ 0.15M⊙ yr−1 kpc−2). As previously, we note thatthe global dust attenuation corrected Hα luminosity appears tooverestimate the global SFR compared to other multiwavelengthmethods (Section 7 and Table 3). In the nuclear regions, wherethe dust attenuation is large, it is possible that the dust atten-uation correction is leading to incorrect emission line strengthestimates.


Table 3.The star formation rates of NGC 4676A and NGC 4676B esti-mated from CALIFA and multiwavelength observations.

Method SFRA/M⊙yr−1 SFRB/M⊙yr−1

(Hα,Hβ)a 17-35 3-5(Hα,mid-IR)b 4.3-6.9 1.3-2.1

stellar continuumc 2.6-5 1-4(FUV,mid-IR)d 6.2 2.3

[Ne ii]e <10.1 <1.4(FIR, radio)f 14 4

Notes. (a) Dust attenuation corrected Hα luminosity using the Balmerdecrement (Section 6.1). The errors on the dust attenuation correctionin the nucleus of NGC 4676A are likely to be significant, due to thevery high dust content.(b) Combination of Hα and 24µm luminosities(Smith et al. 2007).(c) Decomposition of the stellar continuum usingstarlight (Section 5.3).(d) Combination of 24µm and FUV luminosities(Smith et al. 2010).(e) Mid-IR [Ne ii] emission line (Haan et al. 2011).(f) Combined FIR and radio continuum luminosities (Yun & Hibbard2001).

7. Multiwavelength analysis of global star formationrates

In the previous sections we have estimated the ongoing starformation rate of the Mice galaxies from both the ionised gasrecombination lines and the stellar continuum in the CALIFAdatacube, however, both estimates are uncertain. Althoughthehigher SFR for NGC 4676A estimated from the emission linescompared to the stellar continuum may plausibly indicate a re-cent increase in SFR, we urge caution in this interpretation. Asignificant fraction of the line emission may arise from gas thathas not been photoionised by stars, and the very large dust at-tenuation causes significant uncertainty in the central region ofNGC 4676A. The SFR obtained from decomposition of the stel-lar continuum depends sensitively on the star formation historyfitted by thestarlight code. While both methods suggest thatonly moderate levels of star formation is ongoing in both galax-ies, there are several multiwavelength observations that can beused to verify this result. These are summarised in Table 3 anddiscussed in more detail below.

The Mice have been observed in the mid-IR with the MIPSinstrument on board Spitzer (Smith et al. 2007) and the NUV andFUV with GALEX (Smith et al. 2010). This provides us withtwo alternative SFR estimates which account for optically thickdust obscuration by direct detection of the thermal dust emis-sion. Firstly, we combine the 24µm luminosity with the totalHα line luminosity, using both the Calzetti et al. (2007) coeffi-cient of 0.031 derived from Hii regions and the Kennicutt et al.(2009) coefficient of 0.02 which includes a correction for diffuseemission. We note that this calibration depends on metallicity(e.g. Relaño et al. 2007), but is appropriate for the solar metal-licity that we measure for the Mice galaxies in Section 6.2. Thismethod results in a SFR of 4.3-6.9 M⊙yr−1 for NGC 4676A and1.3-2.1 M⊙yr−1 for NGC 4676B, where the ranges account forthe different coefficients and the upper and lower limits on theHα line flux that arises from Hii regions.

We can combine the 24µm with the FUV luminosities to ob-tain a SFR independent of the CALIFA data, along with an atten-uation in the FUV (AFUV), following the prescription in Iglesias-Páramo et al. (2006)4. This results in a SFR of 6.2 M⊙yr−1,with an attenuation ofAFUV = 3.3 mag for NGC 4676A, and

4 FUV luminosities are converted into SFRs using SB99 (Leithereret al. 1999) and a Salpeter IMF withMlow = 0.1 M⊙ andMup = 100 M⊙.

SFR of 2.3 M⊙yr−1 with an attenuation ofAFUV = 1.3 mag forNGC 4676B.

We can use the [Neii]12.81µm flux to provide a nebular lineestimate of SFR that is unaffected by dust obscuration. Takingthe [Neii] flux reported in Haan et al. (2011) and the calibrationof Diamond-Stanic et al. (2012) for galaxies withLIR < 1011L⊙,gives a SFR of<10.1 and<1.4 M⊙yr−1, for NGC 4676A and Brespectively. The upper limits arise because, like Hα, [Ne ii] canbe emitted by processes other than star-formation.

Finally, Yun & Hibbard (2001) combine FIR IRAS observa-tions with radio continuum flux to estimate the FIR luminosityof each galaxy, even though they are not spatially resolved inIRAS. Assuming a factor of 2.5 conversion from Helou et al.(1988) to total infrared flux, and Kennicutt (1998b) conversionto SFR we obtain a SFR of 14 and 4M⊙yr−1, for NGC 4676Aand B respectively. Combining Hα luminosity with our own es-timate of IRAS total infrared flux from template fitting between8-1000µm and the conversion given by Calzetti (2013) results ina lower total SFR for the two galaxies, but consistent withinthescatter of the other measurements.

Given the complexity of the galaxies, it is perhaps surprisingthat the many different estimates of SFR are so close. Excludingthe [Neii] based estimate, for which we only have an upper limit,we obtain median values of 6.2 and 2.3 M⊙yr−1, for NGC 4676Aand B respectively.

Dividing the total SFR of the galaxies by their stellar mass(Section 5.3) gives a specific SFR (sSFR) of 5× 10−11yr−1

and 1.3 × 10−11yr−1 or log(sSFR/yr−1) of −10.3 and−10.9 forNGC 4676A and B respectively.

8. The bicone in NGC 4676A

In Figure 9 we show two emission line diagnostic diagramsfor individual spaxels in NGC 4676A. Spaxels in the disk arecoloured blue and those in the bicones are coloured orange,with lightness of tone increasing with distance from the nucleus.While gas in the northern tidal tail and north and south diskshows line ratios similar to high metallicity star-forminggalax-ies in the local Universe (Kauffmann et al. 2003; Stasinska et al.2006), the bicones have line ratios which indicate increasinghardness of the ionisation field with height above the mid-planeof the disk. At the outer extent the line ratios lie primarilyin theregion commonly occupied by Seyfert galaxies (Cid Fernandeset al. 2010; Kewley et al. 2006). Increasing hardness of the radi-ation field with distance from the mid-plane, and extended softX-ray emission (Read 2003) is inconsistent with an AGN beingthe primary source of ionisation, but instead suggests thatshocksdriven by a superwind are ionising the gas (Heckman et al. 1990,hereafter HAM90).

Overplotted on Figure 9 are predictions from the fast-shockmodels of Allen et al. (2008)5, for a range of shock velocities(vs) and magnetic field strengths (B). We use solar metallicityas measured from line ratios in the disk (Section 6.4), the abun-dance set of Grevesse et al. (2010) and a pre-shock density of1cm−3. In photoionising shocks, the flux of ionising radiationemitted by the shock increases proportional tov3s , leading to astrong increase in pre-ionisation level of the gas as shock veloc-ity increases. At the highest velocities, the ionisation front canexpand ahead of the shock front, leading to an “ionised precur-

24µm luminosities are converted to total IR luminosities (from 8µm to1000µm) using the models by Chary & Elbaz (2001).5 These were calculated using the ITERA package: (Groves & Allen2010).



Kewley et al. (2001)

Kauffmann et al. (2003)

Stasinska et al. (2006)

Cid Fernandes et al. (2010)

Kewley et al. (2006)

Nucleus

Knot A−N/S

"Outflow"

Tail

Other spaxels

BiconeRadial distance from nucleus

Fig. 9. Line ratio diagnostic diagrams, showing line ratios for independent spaxels in NGC 4676A. Those spaxels which lie in the bicones arecoloured orange, shaded according to radial distance from the nucleus, the spaxels lying in the non-shocked regions are included as blue circles.Line ratios in some key regions of interest (see Fig. 2) are plotted as yellowsymbols. Overplotted as black lines are empirically and theoreticallyderived separations between LINERs/Seyferts and Hii regions. Overplotted as coloured lines are line ratios predicted for the photoionisation ofgas by fast shocks from Allen et al. (2008), as described in the text. The blue model tracks show predicted line ratios when ionisation comes fromthe shock front alone, orange lines also include pre-ionisation by a precursor.

sor” which can contribute significantly to the optical emissionof the shock. Magnetic fields can limit compression across theshock, thereby allowing the ionisation front to proceed into thepost-shocked gas more quickly than the case with no magneticfields, again leading to an increase in pre-ionisation levelof thegas. We overplot predicted line ratios from models both withandwithout inclusion of the additional photoionisation caused by theprecursor (orange and blue lines). The observed line ratiosin thecone of NGC 4676A are consistent with a fast shock, includingan ionised precursor, with velocity increasing from∼200 km s−1

in the centre of the galaxy to∼350 km s−1 at the outer edge ofthe bicone. We note that this apparent increase in shock velocitycould also arise from a line-mixing scenario, where the emissionfrom [H ii] regions falls off more rapidly with height above thedisk than emission from shocks. Further data would be neededto confirm whether or not the shock was increasing in velocityfrom the mid plane.

Recent low velocity shock models described in Rich et al.(2010, 2011) and Farage et al. (2010) are appropriate for shockswith velocities. 200 km s−1. As these models are not available,we compare to the figures published in Rich et al. (2011) forNGC 3256, which has a similar metallicity to NGC 4676A. Wefind that the large observed [Oiii]/Hβ line ratio in NGC 4676Ais inconsistent with the slow shock models, and the closest mod-els have velocities of. 100 km s−1, well below the measuredvelocity dispersion of the lines in the bicones.

8.1. Outflow kinematics

Shocks heat the gas through which they pass, and lead to in-creased linewidths of the post-shock emitting gas from thermalmotions. Line splitting of absorption and/or emission lines issometimes observed in galactic outflows, due to the bulk mo-tions of gas towards and away from the observer. An increase inHα emission line width is observed in radial fingers extendingalong the minor axis of NGC 4676A (Figure 3), with a max-

imum velocity dispersion ofσ ∼ 200 km s−1. Unfortunately,with the limited spectral resolution of the CALIFA data we areunable to resolve the different kinematic components, thus wecan only use the measured line widths to place limits on the bulkkinetic motions of the gas. If the increased linewidth is causedby bulk outflow, then the line-of-sight component of the outflowvelocity is estimated from half of the full-width-half-maximum(FWHM) of the line to be∼235km s−1. Including a significanttangental component, as might be expected from an outflowingwind, this is consistent with the estimate of the shock velocityestimated from line ratios above.

8.2. Energetics of the superwind fluid

Spatially resolved, optical line emission observations can con-strain the rates at which the fast moving wind fluid inNGC 4676A carries mass, momentum and energy out of thegalaxy. Within the context of the superwind model of HAM90,the wind in NGC 4676A has expanded beyond the initial “hotbubble” phase, into the “blow out” or free expansion phase, dur-ing which the wind propagates at approximately constant veloc-ity into the intergalactic medium. The optical emission linesarise from clouds and wind shell fragments that are shock heatedby the outflowing wind fluid.

The density of the medium into which the wind is propagat-ing (n1) can be determined from the electron density (ne) mea-sured from the [Sii] line ratios (Section 6.3), combined withknowledge of the type of shock causing the optical line emission(see Appendix A for details):

n1[cm−3] = 0.12

(

ne[cm−3]100

) (

350vs[km s−1]

)2

(1)

wherevs is the shock velocity. From this, the thermal pressureof the clouds where the [Sii] emission arises is given by:

Pcloud = n1mpµ v2s = 6.6× 10−11Nm−2 (2)


wheremp is the proton mass,µ = 1.36 accounts for an assumed10% Helium number fraction and we have setne = 200cm−3

from the observed [Sii] line ratio at the outer extent of the bi-cones. We note that, for fast shocks, the thermal pressure ofthecloud is independent of the shock velocity. This is equivalent to apressure of 6.6×10−10dynes cm−2 or Pcloud/k = 4.8×106K cm−3.The pressure at the outer extent of the wind in NGC 4676Ais a little larger than the range measured by HAM90 for 6 far-infrared galaxies (FIRGs) of 2.5− 5× 10−10dynes cm−2.

In the “blow-out” phase of a superwind, the pressure sourcein the outer regions of the wind is the ram pressure of the windfluid itself 6: Pwind ∼ Pcloud. This allows us to use the informa-tion derived from the shock heated gas to determine the momen-tum flux of the wind flowing into a solid angleΩ. FollowingHAM90:

pwind = Pwind(r)r2 Ω

4π= 1.3× 1030N (3)

wherePwind(r) is the wind pressure at radiusr and we have es-timatedr = 6.6 kpc andΩ/4π ∼ 0.37 from the CALIFA maps.This is equivalent to 1.3 × 1035dynes. This compares to valuesranging from 0.3− 12× 1035dynes for the winds in HAM90, al-though we note that these will be upper limits, as without IFSdata the authors assumedΩ/4π = 1.

Both the energy and mass outflow rates can be estimatedfrom the momentum flux, however, they depend upon the un-known velocity of the wind fluid. For typical values for super-winds ofvw = 1000−3000 km s−1 (Hopkins et al. 2013; Seaquistet al. 1985, HAM90) we estimate a total energy flux of

Ewind = 0.5Pwindr2vwΩ

4π= [6.4− 20.4] × 1042 erg s−1 (4)

and a mass outflow rate of

Mwind =Pwindr2

vw

Ω

4π= [8 − 20]M⊙yr−1. (5)

This mass outflow rate is a factor of 1.5-3 larger than the globalstar formation rate of the galaxy (M∗ ∼ 6 M⊙yr−1, Section 7),which is typical for galaxy outflows (e.g. Martin 1999; Veilleuxet al. 2005).

8.3. The optical emission nebulae

Given the LINER- and Seyfert-like line ratios observed in thebicones, the optical line emission seen at large distances fromthe galaxy must largely be a direct result of the fast, radiativeshocks which are being driven into the ambient medium by thesuperwind.

To obtain the total dust attenuation corrected Hα luminosity7

in the bicones we sum over regions with log([Nii]/Hα)> −0.25and conservatively exclude emission within 1.3 kpc (3′′) of the

6 A lower limit on the ram pressure of the wind fluid can also be es-timated from the X-ray data. Taking the density and temperature fromRead (2003) we findP/k > 1×105K cm−3, consistent with the CALIFAresults. However, we note that this value is highly uncertain: the lackof counts in the X-ray data have deterred other authors from fitting theX-ray spectrum to obtain a temperature (González-Martín et al. 2009).Improved X-ray data would be required to estimate a filling factor forthe X-ray emitting gas.7 Where Hβ is not measured we use uncorrected Hα line luminosities,although this is only the case in the fainter outer regions and makes nodifference to the final results.

nucleus, where the [Nii]/Hα line ratio lies in the composite re-gion of the BPT diagram and we might expect a significant frac-tion of line emission to instead arise from photoionisationbystars in the starburst. We also exclude the highly ionised gasto the north and south of the nucleus, as this may have a dif-ferent origin. We find the total Hα luminosity of the nebulae isLneb = 7 × 1041erg s−1, leading to a bolometric energy loss rateof Eneb ∼ 5.5× 1043erg s−1, where we have assumed a bolomet-ric conversion factor of 80, appropriate for ionisation by shockswith vs & 140 km s−1 (Rich et al. 2010). This is a factor of 2-6 larger than the energy flux in the wind fluid estimated fromthe geometry and cloud pressure (Ewind, Equation 4). Given theuncertainties in the many assumptions made to obtain these twonumbers, we consider them to be in good agreement, implyinga near 100% efficiency in converting wind energy into radiation.However, it is possible that an additional source of ionisationis contributing to the optical nebulae, or our zeroth order cal-culations have underestimated the energy flux of the wind bya factor of a few. In comparison, HAM90 find the total dustattenuation corrected line luminosity in FIRGs to range from1.5 × 1042 − 1.5 × 1044, assuming a bolometric correction of30. They also note thatEwind ≈ Lneb implies a high efficiency forconverting the wind energy into emission lines.

From the optical emission flux we can also estimate the totalmass of ionised gas currently decelerating at the shock fronts:

Mion =µmpLHα,0

γHα(T )ne(6)

where, for Case B recombination and purely photoionised gasof electron temperatureT , the effective volume emissivity isγHα(T ) = 3.56× 10−25T−0.91

4 erg cm−3 s−1 (T4 = T/104K). Tak-ing T = 104 and the median electron density throughout thebicones ofne = 400cm−3 we find a total ionised gas mass of∼ 6× 106M⊙.

For a shock front travelling at 350 km s−1 the crossing timeto reach the outer edge of the bicones is∼18 Myr. Assumingthe shocks cause a bulk motion of the entrained gas at the veloc-ity of the shock, this gives an ionised gas mass outflow rate of0.3 M⊙yr−1. We see that the mass of outflowing ionised gas isnegligible compared to the total mass in the wind (Eqn. 5).

8.4. Comparison to the energy injection rate by SNe

The mechanical energy injection rate into the ISM by supernovae(SNe) and stellar winds can be estimated from evolutionary syn-thesis models of populations of massive stars (Leitherer etal.1999; Veilleux et al. 2005):

E∗ = 7× 1041

(

SFRM⊙yr−1

)

= 4.3× 1042 erg s−1 (7)

for SFR= 6.2M⊙yr−1(Section 7), and we have taken the limit-ing assumption that the mechanical energy from all of the starsbeing formed throughout the entire galaxy is available to drivethe wind. Even under this assumption, this barely provides suf-ficient energy to drive the lowest velocity wind assumed above,and is a factor of 10 too little to power the emission line nebulae.The geometry of the outflow as seen in the CALIFA observa-tions suggests that only the central star formation is driving thewind, increasing the discrepancy. This implies that other formsof energy injection into the biconical nebulae are requiredbe-yond simple mechanical energy, such as photo-heating from theyoung stars and radiation pressure (see e.g. Hopkins et al. 2013for recent simulations including these effects).



−1.5 −1.0 −0.5 0.0 0.5log([N II] λ6583/Hα)

−1

0

1lo

g([O

III] λ5

007/

Hβ)

log u = −4.0

−3.6

−2.6

−2.0α = −1.2

−1.7

−2.0

Dusty AGN photoionization(Groves et al. 2004)

Seyfert

LINER

NGC 4676 B

Other spaxels

Bicone

Nucleus

Knot B

−2.0 −1.5 −1.0 −0.5 0.0log([O I] λ6300/Hα)

−1.5

−1.0

−0.5

0.0

0.5

1.0

1.5

log(

[O III

] λ500

7/H

β)

log u = −4.0

−3.6

−3.0

−2.6

−2.0α = −1.4−1.7

−2.0

Seyfert

LINER

Fig. 10. Line ratio diagnostic diagrams, showing line ratios for independent spaxels in NGC 4676B. Those spaxels which lie in the bicones arecoloured orange, the remainder of the spaxels are included as blue circles. Line ratios in some key regions of interest are shown as yellow symbols(see Fig. 2). Overplotted as black lines are empirically and theoretically derived separations between LINERs/Seyferts and Hii regions, as givenin Figure 9. Overplotted as coloured lines are AGN photoionisation model predictions by Groves et al. (2004).

9. The bicone in NGC 4676B

A bicone is also evident in the line ratio maps of NGC 4676B(Figure 7). The different pattern of line ratios to those seen inNGC 4676A suggest that different physical mechanisms are re-sponsible. In Figure 10 we plot two emission line diagnosticdiagrams for individual spaxels in NGC 4676B. As seen in themap, the bicone does not have distinct [Oiii]/Hβ line ratios com-pared to the disk. The nucleus has line ratios in the “composite”region in [N ii]/Hα, and in the LINER region in [Oi]/Hα.

The hard X-ray detection, compact distribution of soft X-rays and unusually high ratio of mid-IR excited H2 emission toPAH emission suggest an AGN may be present in NGC 4676B(González-Martín et al. 2009; Masegosa et al. 2011; Read 2003,see Section 1). If so, the AGN is weak, with a hard X-ray lu-minosity ofLX,2−10keV = 1.48× 1040erg s−1. Using a bolometricconversion factor of 50 suitable for LINERs (Eracleous et al.2010b), results in a bolometric luminosity of 7.4 × 1041erg s−1

or log(L/L⊙) = 8.3, typical for LINER galaxies. From theCALIFA nuclear spectrum we measure an [Oiii] luminosity oflog(L[OIII] /L⊙) = 5.9. Using a bolometric conversion factor of600 (Heckman et al. 2004), leads to a bolometric luminosity of1.7 × 1042erg s−1 or log(L/L⊙) = 8.7. Given the errors inherentin these conversions, these values are consistent.

Predicted line ratios from the AGN photoionisation modelsof Groves et al. (2004) are overplotted in Figure 10, for a rangeof dimensionless ionisation parameter (u ≡ q/c) and spectral in-dex (α). The line ratios in the nucleus are consistent with anAGN with logu ∼ −3.3 andα ∼ −1.5, with line mixing from[H ii] regions in the host galaxy disk causing the variation acrossthe galaxy. It is also worth considering the potential of post-AGB stars in the underlying old stellar population to contributeto the ionising photon field, particularly in the massive bulgeof NGC 4676B (see e.g. Eracleous et al. 2010a). However, theobserved Hα equivalent width in the nucleus of NGC 4676B is∼ 25Å, which firmly rules out any significant contribution tothe line emission from post-AGB stars (Cid Fernandes et al.2011). Finally, the nuclear line ratios are also consistentwithslow-shock models including some line mixing from low levelstar formation (e.g. Rich et al. 2011). Ultimately, understand-ing fully the source of the high ionisation lines in NGC 4676B

0.0 0.5 1.0 1.5 2.0 2.5Time (Gyr)

0

10

20

30

40

50S

FR

(M

Ο •/yr)

Fig. 11. The combined SFR of the mock Mice over the duration of themerger simulation. The blue and red vertical marks indicate the time offirst passage and time of observation at 180 Myr after first passage.

will require observations with higher SNR, spatial and spectralresolution.

10. Comparison to simulations

Since the initial galaxy merger simulations by Toomre & Toomre(1972), a large research effort has focussed upon trying to re-produce the spatial distribution of stars, gas, star formation andstellar cluster ages in mergers (e.g. Karl et al. 2010; Lanz et al.2014). With their tidal tails constraining the orbital parame-ters reasonably well, the Mice have been an obvious target forsuch studies. For example, Mihos et al. (1993) found the SFRof NGC 4676A to be higher in models than data, which couldresult from limitations of the data available at the time (e.g. in-ability to correct for dust attenuation) or limitations of the model.Barnes (2004) used the Mice as a case study to advocate a shock-induced star formation law, rather than the standard gas densitydriven star formation implemented in hydrodynamic simulations(see also Teyssier et al. 2010). Recently, Privon et al. (2013) pre-sented a new solution for the orbital parameters of the Mice frompurely N-body simulations, which differ from those of Barnes(2004), largely due to the use of a different mass model for thegalaxies. The degeneracy between mass model and orbital pa-


20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

ΣV,corr

23.0

21.5

20.0

18.5

17.0

N

E

-40 -20 0 20 40 60∆α (arcsec)

0

50

100

150

∆δ (

arcs

ec)

ΣV

23.0

21.5

20.0

18.5

17.0

20 0 -20∆α (arcsec)

-20

0

20

40

60

80

∆δ (

arcs

ec)

log( ΣSFR)

-4.0

-3.0

-2.0

-1.0

0.0

N

E

-40 -20 0 20 40 60∆α (arcsec)

0

50

100

150∆δ

(ar

csec

)log( ΣSFR, Starlight )

-4.0

-3.0

-2.0

-1.0

0.0

Fig. 12. Top: The V-band surface brightness (mag/′′) of the real (left) and mock (right) Mice. The surface brightness of the real Mice has beencorrected for dust attenuation using the analysis from thestarlight code (Section 5). The surface brightness of the mock Mice has been calculatedby post-processing an SPH simulation to obtain a mock IFU datacube with the same spectral range and resolution as the CALIFA data. This hasbeen analysed using the same pipelines as the real data.Bottom: The star formation rate surface density (M⊙ yr−1 kpc−2) measured from the stellarcontinuum using thestarlight code, averaged over the last 140 Myr, for the real (left) and mock (right) Mice. White indicates regions with nomeasurable star formation.

rameters are discussed in detail in Privon et al. (2013) and inhibiteasy comparison between simulations and observations.

With observations advancing towards an era in which spa-tially resolved star formation histories, kinematics, ionisationmechanisms and AGN strengths are well constrained for singleobjects, simulations are facing more stringent observational con-straints than ever before (e.g. Kronberger et al. 2007). Given theprimarily observational nature of this paper, we present here anillustrative, rather than exhaustive, comparison of the CALIFAresults with a standard smoothed particle hydrodynamic (SPH)merger simulation. It is simulations such as these that havedriven many of the currently favoured models of galaxy evolu-tion over the last decade, and it is worth investigating how wellthey match the latest generation of observations. For concise-ness, we focus on the stellar kinematics, SFR surface densities

and star formation histories of the galaxies, leaving a morecom-plete study for further work.

The simulation represents the collision of two galaxies witha baryonic mass ratio of 1.3: a pure exponential disk galaxyand a galaxy with a bulge-to-disk ratio of 0.27. These pa-rameters are very close to the observed mass ratio and mor-phologies of the Mice galaxies. We take the orbital parametersprovided in Barnes (2004), i.e. a close to prograde-progradeorbit with a closest approach of∼10 kpc occurring∼170 Myrago. Star formation and the associated supernova feedback isimplemented using the sub-resolution multiphase model devel-oped by Springel & Hernquist (2003), in which cold gas formsstars when its density reaches above a certain density threshold(nH = 0.128 cm−3, appropriate for the resolution of the simula-tion). For the purposes of this CALIFA comparison, black holefeedback is not included.



-40 -20 0 20 40 60∆α (arcsec)

0

50

100

150

∆δ (

arcs

ec)

∆ v

-300

-200

-100

0

100

200

-40 -20 0 20 40 60∆α (arcsec)

0

50

100

150

∆δ (

arcs

ec)

σv

0

55

110

165

220

Fig. 13. The stellar velocity field and dispersion of the mock Mice. See Figure 3 to compare with the data.

Spectral energy distributions of each of the star particleswere calculated from their star formation history, using the mod-els of Bruzual & Charlot (2003). In order to make a fair com-parison between observations and simulations, we convert thesimulation into an IFS data cube with the same spectral and spa-tial resolution as the data, and analyse this cube using the samecodes as employed on the real data. Full details of the simula-tions and post-processing applied to produce the mock observa-tions are given in Appendix B.

Figure 11 presents the global SFR of both galaxies in thesimulation, over the full 3 Gyr run-time. The time of first pas-sage and time of observation are marked as vertical lines. Thesmall burst of star formation caused by first passage is evident.At final coalescence the simulations predict a 40-50M⊙yr−1 star-burst; however, more accurate gas mass fractions of the realMicewould be required to improve the accuracy of this prediction.

The upper panels of Figure 12 show maps of theV-band sur-face brightness of the real and mock Mice. The data has beencorrected for dust extinction using theV-band stellar continuumattenuation map estimated by thestarlight code. On altering theviewing angle of the simulation a strong bar becomes visibleinNGC 4676A. This lends support to our argument that the boxyshape of NGC 4676A is caused by an edge-on bar rather than aclassical bulge (Section 4.1). While the sizes and overall surfacebrightnesses of the real and mock galaxies are largely compa-rable, the mock galaxies are separated by a distance on the skythat is a factor of 2 larger than observed. This may result from in-correct initial orientations, orbits, mass profiles, or shapes of thedark matter halo. The new orbital parameters presented in Privonet al. (2013) differ in several aspects to those used here. How-ever, because these authors assume very different mass modelsfor the progenitor galaxies, their orbital parameters are no morelikely to provide a better match. Simply slowing the initialve-locities of the simulated galaxies resulted in too short tidal tails,and further investigations into the origin of the discrepancy arebeyond the scope of this primarily observational paper.

The lower panels of Figure 12 show the SFR surface den-sity averaged over the last 140 Myr, calculated for both the realand mock Mice from the decomposition of the stellar contin-uum by thestarlight code (Section 5). The SFR surface den-

sity reconstructed from the mock IFS data cube usingstarlightclosely matches the instantaneous SFR surface density taken di-rectly from the simulation. The global SFR of each of the mockgalaxies is∼ 2.5M⊙yr−1, in agreement with the observed SFRfor NGC 4676B (2M⊙yr−1) and a little lower than observed forNGC 4676A (6M⊙yr−1).

The SFR surface densities in the central 5× 5′′ (2.2 kpc)are∼ 0.3M⊙ yr−1 kpc−2 for both mock galaxies. The value av-eraged over 140 Myr using thestarlight spectral decomposi-tion is close to the instantaneous value obtained directly fromthe simulation. This compares well to a central SFR densityof 0.15M⊙ yr−1 kpc−2 for NGC 4676A and is not substantiallydifferent from the∼ 0M⊙ yr−1 kpc−2 measured for NGC 4676Bfrom the starlight spectral decomposition (Section 5). How-ever, the observed decrease in SFR surface density towards thecentral regions of NGC 4676B is not reproduced by the simula-tions.

Figure 13 shows the stellar velocity field and velocity disper-sion maps of the mock Mice. The velocity map of NGC 4676Ais a good match to the data, and including a bulge results in asignificantly worse match. The velocity map of NGC 4676B isdominated by a rotating disk, with the bulge causing the pinchedeffect at the centre and small rise in velocity dispersion. The un-usual twisted stellar disk velocity field seen in the observations,with the rotation axis offset from the minor axis, is not observedat this stage of the simulation. During the full run time of thesimulation, the high surface brightness of the tidal tails in theMice is clearly a transient feature. There is also evidence for aslight twist in the velocity field of NGC 4676B at some epochsof the simulations, similar to that seen in the observations. Pre-cisely reproducing these transient morphological and kinematicfeatures may make it possible to set tighter constraints on the or-bital parameters of the Mice merger and mass distribution ofthegalaxies in the future.

11. Discussion and Summary

Here we summarise the impact of the close encounter on themorphology, kinematics, star formation rate and history, and


Table 4. A summary of the main properties of the Mice galaxies mea-sured in this paper. For a summary of properties available from theliterature see Table 1.

Parameter NGC 4676A (NW) NGC 4676B (SE)B/Ta 0 (SBd) 0.5 (S0/a)

vnucleus,gas (km/s)b 6585 6581vnucleus,stars(km/s)b 6652 6493

M∗ (1011M⊙)c 1.2 1.5SFR (M⊙/yr)d ∼6 ∼2

P/kwind (106K cm−3)e 4.8 ...Mout (M⊙yr−1)e 8-20 ...

Morphological PAf 2.8 33.2Kinematic PAstars

f 33.9, 27.6 165.4, 150.4Kinematic PAgas

f 31.9, 24.5 154.4, 165.0

Notes.(a) Bulge-to-total flux ratio in theI-band, from image decompo-sition of HST-ACS image (Section 4.1).(b) Recessional velocity, fromkinematic fits to the central spaxel with the highestV-band luminos-ity (Section 4).(c) Total stellar mass, from full spectrum decompositionassuming a Salpeter IMF (Section 5).(d) Global SFR taken from themedian of CALIFA and multi-wavelength measurements (Section 7).(e) The ram pressure and mass outflow rate of the bi-conical wind fromNGC 4676A (Section 8).(f) Position angles (approaching and recedingfor kinematic PAs), measured within 10′′ (Barrera-Ballesteros et al. inprep., Section 4).

ionised gas of the merging galaxies. Table 4 collates some ofthe key properties measured in this paper.

1. Merger induced bars in NGC 4676B and NGC 4676A?The strong bar in NGC 4676B is clear, both in the image andkinematics maps: the characteristic Z-shaped iso-velocitycontours are visible in both the ionised gas and, unusually,in the stellar velocity field. While NGC 4676A has previ-ously been classified as an S0, the CALIFA maps of the stel-lar kinematics show no evidence for a classical bulge, withlow nuclear velocity dispersion and constant rotation abovewith height above the major axis. A young, thin, disk is visi-ble in the stellar population maps, with older populations ex-tending into the “boxy” shaped bulge. The high dust and gascontent of the galaxy is also inconsistent with its classifica-tion as an S0. It is possible that the boxy morphology evidentin the imaging (See Figure 1) comes from a strong bar (e.g.Bureau & Athanassoula 1999, 2005; Kuijken & Merrifield1995; Martínez-Valpuesta et al. 2006; Williams et al. 2011);further detailed investigation of higher order moments in thevelocity map may help to confirm this. The first passage ofa major merger is seen to induce strong bars in simulationsof disky galaxies (e.g. Barnes & Hernquist 1991; Lang et al.2014), including the mock Mice simulation presented here.While we cannot rule out the pre-existence of bars in themerging galaxies, the presence of such strong bars in bothgalaxies, combined with predictions from simulations, sug-gests that these have been induced by the recent close en-counter.

2. Z-shaped stellar velocity field in NGC 4676B.The CAL-IFA data reveals that the dominant rotation of both theionised gas and stars in NGC 4676B is close to being aroundthe major axis of the galaxy (offset between morphologicaland kinematic PAs of 50-60 degrees). While these Z-shaped(or S-shaped) isovelocity contours are a common feature inthe gas velocity fields of strongly barred galaxies (e.g. Con-topoulos & Papayannopoulos 1980), and weak warps can be

seen in the outer regions of stellar disks (Saha et al. 2009),such strong disturbance of the inner stellar velocity field isunusual (for one other example see NGC 4064, Cortés et al.2006). Barrera-Ballesteros et al. (2014) have confirmed thatbars have no significant effect on stellar velocity fields in thefull CALIFA sample. Studies of larger samples are neededto see whether such strong twists are common in merginggalaxies (see Barrera-Ballesteros et al, in prep., for suchastudy within CALIFA).

3. No substantial increase in global star formation rates.Wecan compare the sSFR of the Mice galaxies to the distribu-tion of aperture corrected sSFR as a function of stellar massfor galaxies in the SDSS survey (Figure 24 of Brinchmannet al. 2004). We find that NGC 4676A lies at the upper endof stellar masses measured for star-forming galaxies in thelocal Universe and has exactly typical sSFR for its stellarmass. NGC 4676B has a slightly low sSFR for its stellarmass, but still lies well within the range observed in the gen-eral population. Both galaxies have SFR surface densitiestypical of star-forming disk galaxies. We conclude that thereis no evidence that either galaxy is currently undergoing asubstantial galaxy-wide burst of star formation. These re-sults are broadly consistent with the mock Mice simulations,where the galaxies have only undergone a small increasein global SFR since first passage. Circumstantial evidencefor increased star formation in close pairs has existed fordecades (e.g. Bernloehr 1993). Statistically significant evi-dence for the enhancement in star formation due to interac-tions and mergers has come from large samples of spectro-scopically observed galaxies in the last decade (Barton et al.2000; Lambas et al. 2003; Patton et al. 2011, 2013; Woodset al. 2010). Estimates of the occurrence, strength and du-ration of SFR enhancement vary. For example, Woods et al.(2010) find that SFR increases by a modest factor of 2-3 aver-aged over the duration of the enhancement, which is arounda few hundred million years. Such enhancements occur inabout 35% of close pairs. These observational results areconsistent with hydrodynamic galaxy merger models (Mat-teo et al. 2008) and show that the lack of SFR enhancementin the Mice galaxies is not unexpected.

4. No substantial post-starburst population. First passageof the Mice merger occurred roughly 170 Myr ago (Barnes2004; Chien et al. 2007) and star formation induced at thisepoch should be identifiable through decomposition of thestellar continuum to measure the fraction of light from Aand F stars. Referring to the fraction of light emitted by starsformed between 140 Myr and 1.4 Gyr ago in Figure 5, wesee that during this time star formation is located primarilyin the disks of both galaxies, as expected for continuous starformation histories. There is some enhancement in the inter-mediate age population in the outer extents of the galaxies.Our results show that, while some low level star formation atlarge radii may have been triggered by the first passage, thefraction of total stellar mass formed was not significan.

5. Weak nuclear starburst in NGC 4676A.Due to the atten-uation by dust in the centre of NGC 4676A, and poor spatialresolution of far infrared observations, our estimates of thenuclear SFR are uncertain. The SFR surface density in thecentral∼5′′ estimated from thestarlight spectral decompo-sition is∼0.15M⊙ yr−1 kpc−2, placing it at the upper end ofthe range for local spiral galaxies (Kennicutt 1998a). Thisisin good agreement with the mock Mice simulation.

6. No ongoing star formation in the centre of NGC 4676B.Our stellar continuum decomposition of NGC 4676B shows



a complete absence of young stars within∼1 kpc of the nu-cleus (Figure 5). The upper limit on the SFR surface den-sity from the dust attenuation corrected Hα luminosity isΣS FR < 0.2M⊙ yr−1 kpc−2 within the central 5′′. It is clearthat NGC 4676B is not currently undergoing or has not re-cently undergone a nuclear starburst. The strong bar has notyet driven significant gas into the nucleus. This decrease inSFR surface density in the central regions of the galaxy isnot reproduced by the mock Mice simulation.

7. Spectacular bicones driven by fast shocks in NGC 4676A.The extreme line ratios seen at the outer edge of the bi-cones are consistent with being caused by fast shocks(vs ∼ 350 km s−1) driven by a superwind. The emissionof NGC 4676A in soft X-rays is also found to be elon-gated along the minor axis of the galaxy, coincident with theionised gas bicones seen here (Read 2003), again implying astrong galactic outflow. In the nearby Universe observationalstudies of galactic superwinds have focussed primarily onLIRGs and ULIRGs, where a clear driving source is present(see Veilleux et al. 2005, for a review). As NGC 4676A isless luminous than a LIRG, and has a moderate star forma-tion rate despite its very large mass, it would not have beenan obvious candidate for superwind investigations. Stack-ing of rest-frame optical and UV spectra finds evidence forwinds in a large fraction of star forming galaxies at all red-shifts (e.g. Chen et al. 2010; Weiner et al. 2009), implyingthat our census of superwinds is still far from complete andNGC 4676A may not be so unusual. Many more outflowsmay be found in ongoing and upcoming IFU galaxy sur-veys (e.g. Fogarty et al. 2012, in commissioning data for theSAMI survey).

8. What drives the outflow in NGC 4676A?Observational re-sults suggest a minimum SFR surface density of∼ ΣS FR =

0.1M⊙ yr−1 kpc−2 is required for superwinds to be launchedfrom galaxies in the local and distant Universe (Heckman2002, and see the recent compilation by Diamond-Stanicet al. 2012). Thus, although the nuclear starburst inNGC 4676A is not significant in terms of global SFR and isweak with respect to local starbursts, it is powerful enoughto launch a wind in principle. The mass outflow rate fromNGC 4676A is about 1.5-3 times the global SFR of thegalaxy, which is typical for galactic outflows in the localUniverse (Martin 1999). On the limiting assumption thatthe star formation in the entire galaxy is driving the super-wind, we show that the mechanical energy available fromSNe and stellar winds is a factor of 10 too low to explainthe optical line emission in the bicones above 1.3 kpc fromthe plane of the galaxy, and only sufficient to explain the en-ergy outflow rate if the velocity of the wind fluid is very low(< 1000 km s−1). If only the nuclear star formation is drivingthe wind, as suggested by the CALIFA maps, the discrepancyincreases. Additional sources of energy input could be fromionising photons and radiation feedback (Agertz et al. 2013;Hopkins et al. 2013; Murray et al. 2011). Alternatively, theadditional energy could have been provided by a source thathas since switched-off, such as an AGN or a more intensestarburst at first passage.

9. Extended narrow line region (ENLR) in NGC 4676B.Thekiloparsec scale bicones in NGC 4676B are not accompa-nied by higher velocity dispersion gas, and the line ratiosare consistent with ionisation by an AGN. The detection ofhard X-rays in the nucleus and point source like distribu-tion of soft X-rays suggests the presence of a weak AGN.We conclude that there is no evidence for a galactic outflow

from NGC 4676B (as seen in e.g. Cen A, Sharp & Bland-Hawthorn 2010), and the bicones are more likely caused bythe excitation of off-planer gas by an AGN (as seen in e.g.NGC 5252, Morse et al. 1998). Kiloparsec scale ENLRs arenot unusual in local AGN (Gerssen et al. 2012).

10. Areas for improvement in the simulations. The new con-straints afforded by the CALIFA data of the Mice galaxiesindicate new directions where improvements can be made inthe simulations. Firstly, there is some tension between thegeometry of the tidal tails and the relative position of the twoMice, with the simulation Mice flying twice as far apart toobtain the same length and surface brightness of the tails.Further useful constraints on the orbital parameters of themerger and mass profiles of the galaxies may be obtained bydetailed matching of transient morphological and kinematicfeatures, such as the twist in the disk of NGC 4676B, the ori-entation of the two bars and the surface brightness of the tidaltails. Secondly, observations of the gas mass fraction of thegalaxies could be used to further test the ability of the modelsto reproduce the SFR surface density maps, and in particularunderstand the very low star formation rate observed in thenucleus of NGC 4676B.

12. The future is a LIRG... and then?

While the combination of CALIFA and multiwavelength datahas conclusively shown that first passage has not triggered sub-stantial star formation, the considerable gas content of the twogalaxies makes it likely that final coalescence will cause a sub-stantial starburst. The two galaxies have a combined molecu-lar gas mass of 7.2 × 109M⊙ (Yun & Hibbard 2001), which onconversion to stars will lead to a SFR of order 100 M⊙yr−1 overa timescale of 107years and efficiency of 10% (or timescale of108years and efficiency of 100%), values typical for starburstgalaxies (Kennicutt 1998a). This would result in an 8-1000µmluminosity ofLIR ∼ 5×1011L⊙, i.e. a LIRG. The simulation pre-sented in Section 10 predicts a similar peak SFR of∼50M⊙yr−1.

On coalescence the galaxy will have a stellar mass of∼

3×1011M⊙ (Salpeter IMF), assuming that each galaxy increasesits mass by 10% during the starburst as found observationally(Robaina et al. 2009; Wild et al. 2009). Converting to a ChabrierIMF (1.8× 1011M⊙) and comparing to the galaxy mass functionof Baldry et al. (2012) we see that this places the merger remnantin the very highest mass galaxy population in the local Universe.Galaxies of this mass and above have a number density of a few×10−4 Mpc−3.

Extending the predictions for the properties of the systembeyond coalescence of the galaxies requires input from simula-tions. By the end of the hydrodynamic merger simulation pre-sented in Section 10, in about 2.2 Gyr time, the merger remnantwill have a SFR of 1M⊙yr−1, with the decay in star formationoccurring as a natural result of gas consumption. This wouldlead to an elliptical galaxy with a specific star formation rate(SFR/M∗) of 5×10−12 yr−1, within the range measured for galax-ies of this mass in the local Universe (Schiminovich et al. 2007).

Acknowledgements

This study makes use of the data provided by theCalar Alto Legacy Integral Field Area (CALIFA) survey(http://www.califa.caha.es). Based on observations collectedat the Centro Astronòmico Hispano Alemán (CAHA) at CalarAlto, operated jointly by the Max-Planck-Institut für As-tronomie and the Instituto de Astrofisica de Andalucia (CSIC).


CALIFA is the first legacy survey being performed at CalarAlto. The CALIFA collaboration would like to thank the IAA-CSIC and MPIA-MPG as major partners of the observatory, andCAHA itself, for the unique access to telescope time and supportin manpower and infrastructures. The CALIFA collaborationthanks also the CAHA staff for the dedication to this project.

The authors would like to thank the anonymous referee forcomments that significantly improved the paper; Dimitri Gadottifor providing extensive help and advice with the image decom-position; Mike Dopita and Tim Heckman for their patience ex-plaining the effects of shocks in superwinds; Daria Dubinovskafor creating undistorted ACS PSF images from the Tiny Tim im-ages and Carolin Villforth for further help with the ACS PSF;LiaAthanassoula for help interpeting the kinematic maps; DanielPomarede for help getting SDvision running; Jeremy Sandersand Roderik Johnstone for help interpreting the X-ray observa-tions; Eva Schinnerer for pointing out the effect of beam smear-ing; all other interested researchers who have contributedwithquestions and comments following discussions and presenta-tions of this work over the last 2 years.

The numerical simulations were performed on facilitieshosted by the CSC -IT Center for Science in Espoo, Finland,which are financed by the Finnish ministry of education.

Funding and financial support acknowledgements: V. W.from the European Research Council Starting Grant (P.I. WildSEDmorph), European Research Council Advanced Grant (P.I.J. Dunlop) and Marie Curie Career Reintegration Grant (P.I.Wild Phiz-ev); J. M. A. from the European Research Coun-cil Starting Grant (P.I. Wild SEDmorph); F. F. R. O. from theMexican National Council for Science and Technology (CONA-CYT); A. G. from the European Union Seventh Framework Pro-gramme (FP7/2007-2013) under grant agreement n. 267251.The Dark Cosmology Centre is funded by the Danish NationalResearch Foundation. P. H. J. from the Research Funds ofthe University of Helsinki; J. F. B. from the Ramón y CajalProgram, grants AYA2010-21322-C03-02 and AIB-2010-DE-00227 from the Spanish Ministry of Economy and Competitive-ness (MINECO), as well as from the FP7 Marie Curie Actionsof the European Commission, via the Initial Training NetworkDAGAL under REA grant agreement no. 289313; R. G. D. andR. G. B. from the Spanish project AYA2010-15081; A. M.-I.from the Agence Nationale de la Recherche through the STIL-ISM project (ANR-12-BS05-0016-02) and from BMBF throughthe Erasmus-F project (grant number 05 A12BA1). R. A. M.from the Spanish programme of International Campus of Excel-lence Moncloa (CEI); K. J. from the Emmy Noether-Programmeof the German Science Foundation (DFG) under grant Ja 1114/3-2; P. P. from a Ciencia 2008 contract, funded by FCT/MCTES(Portugal) and POPH/FSE (EC); I. M. P. from Spanish grantAYA2010-15169 and the Junta de Andalucia through TIC-114and the Excellence Project P08-TIC-03531; J. M. G. from grantSFRH/B PD/66958/2009 from FCT (Portugal). C. J. W. from theMarie Curie Career Integration Grant 303912; J. I. P. and J. V. M.from the Spanish MINECO under grant AYA2010-21887-C04-01, and from Junta de Andalucía Excellence Project PEX2011-FQM7058; E. M. Q. from the European Research Council viathe award of a Consolidator Grant (PI McLure); M. P. from theMarie Curie Career Reintegration Grant (P.I. Wild Phiz-ev).This work was supported in part by the National Science Foun-dation under Grant No. PHYS-1066293 and the hospitality ofthe Aspen Center for Physics.

ReferencesAgertz, O., Kravtsov, A. V., Leitner, S. N., & Gnedin, N. Y. 2013, ApJ, 770, 25Aguerri, J. A. L., Méndez-Abreu, J., & Corsini, E. M. 2009, A&A, 495, 491Allen, M. G., Groves, B. A., Dopita, M. A., Sutherland, R. S.,& Kewley, L. J.

2008, ApJS, 178, 20Alonso-Herrero, A., Rosales-Ortega, F. F., Sánchez, S. F.,et al. 2012, MNRAS,

425, L46Baldry, I. K., Driver, S. P., Loveday, J., et al. 2012, MNRAS,421, 621Barnes, J. E. 2004, MNRAS, 350, 798Barnes, J. E. & Hernquist, L. E. 1991, ApJ, 370, L65Barrera-Ballesteros, J. K., Falcón-Barroso, J., García-Lorenzo, B., & et al. 2014,

A&A accepted (arXiv:1405.5222), astro-ph.GA, accepted forpublication inA&A

Barton, E. J., Geller, M. J., & Kenyon, S. J. 2000, ApJ, 530, 660Bernloehr, K. 1993, A&A, 268, 25Boquien, M., Duc, P.-A., Galliano, F., et al. 2010, AJ, 140, 2124Brinchmann, J., Charlot, S., White, S. D. M., et al. 2004, MNRAS, 351, 1151Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000Burbidge, E. M. & Burbidge, G. R. 1961, ApJ, 133, 726Bureau, M. & Athanassoula, E. 1999, ApJ, 522, 686Bureau, M. & Athanassoula, E. 2005, ApJ, 626, 159Calzetti, D. 2013, Secular Evolution of Galaxies, 419Calzetti, D., Kennicutt, R. C., Engelbracht, C. W., et al. 2007, ApJ, 666, 870Cappellari, M. & Copin, Y. 2003, MNRAS, 342, 345Cappellari, M. & Emsellem, E. 2004, PASP, 116, 138Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245Casoli, F., Boisse, P., Combes, F., & Dupraz, C. 1991, A&A, 249,359Chary, R. & Elbaz, D. 2001, ApJ, 556, 562Chen, Y.-M., Tremonti, C. A., Heckman, T. M., et al. 2010, AJ, 140, 445Chien, L.-H., Barnes, J. E., Kewley, L. J., & Chambers, K. C. 2007, ApJ, 660,

L105Cid Fernandes, R., Delgado, R. M. G., Benito, R. G., et al. 2014, A&A, 561, 130Cid Fernandes, R., Mateus, A., Sodré, L., Stasinska, G., & Gomes, J. M. 2005,

MNRAS, 358, 363Cid Fernandes, R., Pérez, E., Benito, R. G., et al. 2013, A&A, 557, 86Cid Fernandes, R., Stasinska, G., Mateus, A., & Asari, N. V. 2011, MNRAS,

413, 1687Cid Fernandes, R., Stasinska, G., Schlickmann, M. S., et al. 2010, MNRAS, 403,

1036Conroy, C., Gunn, J. E., & White, M. 2009, ApJ, 699, 486Contopoulos, G. & Papayannopoulos, T. 1980, A&A, 92, 33Cortés, J. R., Kenney, J. D. P., & Hardy, E. 2006, AJ, 131, 747Dale, D. A., Silbermann, N. A., Helou, G., et al. 2000, AJ, 120,583de Vaucouleurs, G., de Vaucouleurs, A., Corwin, H. G., et al.1991, Third Refer-

ence Catalogue of Bright Galaxies, Volume 1-3Diamond-Stanic, A. M., Moustakas, J., Tremonti, C. A., et al. 2012, ApJ, 755,

L26Dopita, M. A. & Sutherland, R. S. 2003, Astrophysics of the diffuse universeEngel, H., Davies, R. I., Genzel, R., et al. 2010, A&A, 524, 56Eracleous, M., Hwang, J. A., & Flohic, H. M. L. G. 2010a, ApJ, 711, 796Eracleous, M., Hwang, J. A., & Flohic, H. M. L. G. 2010b, ApJS,187, 135Farage, C. L., McGregor, P. J., Dopita, M. A., & Bicknell, G. V. 2010, ApJ, 724,

267Fogarty, L. M. R., Bland-Hawthorn, J., Croom, S. M., et al. 2012, ApJ, 761, 169Gadotti, D. A. 2008, MNRAS, 384, 420Gallazzi, A., Charlot, S., Brinchmann, J., White, S. D. M., & Tremonti, C. A.

2005, MNRAS, 362, 41García-Lorenzo, B. 2013, MNRAS, 429, 2903Gerssen, J., Wilman, D. J., & Christensen, L. 2012, MNRAS, 420, 197Girardi, L., Bressan, A., Bertelli, G., & Chiosi, C. 2000, A&AS, 141, 371González Delgado, R., Cerviño, M., Martins, L. P., Leitherer, C., & Hauschildt,

P. H. 2005, MNRAS, 357, 945González-Martín, O., Masegosa, J., Márquez, I., Guainazzi, M., & Jiménez-

Bailón, E. 2009, A&A, 506, 1107Grevesse, N., Asplund, M., Sauval, A. J., & Scott, P. 2010, Astrophysics and

Space Science, 328, 179Groves, B. A. & Allen, M. G. 2010, New Astronomy, 15, 614, (c) 2010 Elsevier

B.V.Groves, B. A., Dopita, M. A., & Sutherland, R. S. 2004, ApJS, 153, 75Haan, S., Armus, L., Laine, S., et al. 2011, ApJS, 197, 27Heckman, T. M. 2002, Extragalactic Gas at Low Redshift, 254, 292, iSBN: 1-

58381-094-3Heckman, T. M., Armus, L., & Miley, G. K. 1990, ApJS, 74, 833Heckman, T. M., Kauffmann, G., Brinchmann, J., et al. 2004, ApJ, 613, 109Helou, G., Khan, I. R., Malek, L., & Boehmer, L. 1988, ApJS, 68,151Hernquist, L. 1990, ApJ, 356, 359Hibbard, J. E. & van Gorkom, J. H. 1996, AJ, 111, 655Hopkins, P. F., Kereš, D., Murray, N., et al. 2013, MNRAS, 433, 78Husemann, B., Jahnke, K., Sánchez, S. F., et al. 2013, A&A, 549,87



Iglesias-Páramo, J., Buat, V., Takeuchi, T. T., et al. 2006, ApJS, 164, 38Johansson, P. H., Burkert, A., & Naab, T. 2009a, ApJ, 707, L184Johansson, P. H., Naab, T., & Burkert, A. 2009b, ApJ, 690, 802Karl, S. J., Naab, T., Johansson, P. H., et al. 2010, ApJ, 715,L88Kauffmann, G., Heckman, T. M., Tremonti, C., & et al. 2003, MNRAS, 346,

1055Keel, W. C., Kennicutt, R. C., Hummel, E., & van der Hulst, J. M. 1985, AJ, 90,

708Kennicutt, R. C. 1998a, ApJ, 498, 541Kennicutt, R. C. 1998b, ARA&A, 36, 189Kennicutt, R. C., Hao, C.-N., Calzetti, D., et al. 2009, ApJ,703, 1672Kewley, L. J., Groves, B., Kauffmann, G., & Heckman, T. 2006, MNRAS, 372,

961Kewley, L. J., Rupke, D., Zahid, H. J., Geller, M. J., & Barton, E. J. 2010, ApJ,

721, L48Kronberger, T., Kapferer, W., Schindler, S., & Ziegler, B. L. 2007, A&A, 473,

761Kubo, J. M., Stebbins, A., Annis, J., et al. 2007, ApJ, 671, 1466Kuijken, K. & Merrifield, M. R. 1995, ApJ, 443, L13Lambas, D. G., Tissera, P. B., Alonso, M. S., & Coldwell, G. 2003, MNRAS,

346, 1189Lang, M., Holley-Bockelmann, K., & Sinha, M. 2014, arXiv:1405.5832, 12

pages, 4 figures, submitted to ApJ LettersLanz, L., Hayward, C. C., Zezas, A., et al. 2014, arXiv:1402.5151Leitherer, C., Schaerer, D., Goldader, J. D., et al. 1999, ApJS, 123, 3Martel, H., Kawata, D., & Ellison, S. L. 2013, MNRAS, 1000Martin, C. L. 1999, ApJ, 513, 156Martínez-Valpuesta, I., Shlosman, I., & Heller, C. 2006, ApJ, 637, 214Martins, L. P., Delgado, R. M. G., Leitherer, C., Cerviño, M., & Hauschildt, P.

2005, MNRAS, 358, 49Masegosa, J., Márquez, I., Ramirez, A., & González-Martín, O. 2011, A&A,

527, 23Matteo, P. D., Bournaud, F., Martig, M., et al. 2008, A&A, 492,31Mihos, J. C., Bothun, G. D., & Richstone, D. O. 1993, ApJ, 418,82Morse, J. A., Cecil, G., Wilson, A. S., & Tsvetanov, Z. I. 1998, ApJ, 505, 159Murray, N., Ménard, B., & Thompson, T. A. 2011, ApJ, 735, 66Navarro, J. F., Frenk, C. S., & White, S. D. M. 1996, ApJ, 462, 563Osterbrock, D. E. & Ferland, G. J. 2006, Astrophysics of gaseous nebulae and

active galactic nucleiPanter, B., Jimenez, R., Heavens, A. F., & Charlot, S. 2007, MNRAS, 378, 1550Patton, D. R., Ellison, S. L., Simard, L., McConnachie, A. W.,& Mendel, J. T.

2011, MNRAS, 412, 591Patton, D. R., Torrey, P., Ellison, S. L., Mendel, J. T., & Scudder, J. M. 2013,

MNRAS, 433, L59Peng, C. Y., Ho, L. C., Impey, C. D., & Rix, H.-W. 2010, AJ, 139, 2097Pettini, M. & Pagel, B. E. J. 2004, MNRAS, 348, L59Pilyugin, L. S., Vílchez, J. M., & Contini, T. 2004, A&A, 425, 849Pomarède, D., Fidaali, Y., Audit, E., et al. 2008, Numerical Modeling of Space

Plasma Flows: Astronum 2007 ASP Conference Series, 385, 327Privon, G. C., Barnes, J. E., Evans, A. S., et al. 2013, ApJ, 771, 120Read, A. M. 2003, MNRAS, 342, 715Relaño, M., Lisenfeld, U., Pérez-González, P. G., Vílchez,J. M., & Battaner, E.

2007, ApJ, 667, L141Rich, J. A., Dopita, M. A., Kewley, L. J., & Rupke, D. S. N. 2010, ApJ, 721, 505Rich, J. A., Kewley, L. J., & Dopita, M. A. 2011, ApJ, 734, 87Robaina, A. R., Bell, E. F., Skelton, R. E., et al. 2009, ApJ, 704, 324Roth, M. M., Kelz, A., Fechner, T., et al. 2005, PASP, 117, 620Roussel, H., Helou, G., Hollenbach, D. J., et al. 2007, ApJ, 669, 959Rupke, D. S. N., Kewley, L. J., & Barnes, J. E. 2010a, ApJ, 710,L156Rupke, D. S. N., Kewley, L. J., & Chien, L.-H. 2010b, ApJ, 723,1255Saha, K., de Jong, R., & Holwerda, B. 2009, MNRAS, 396, 409Sánchez, S. F., Cardiel, N., Verheijen, M. A. W., et al. 2007,A&A, 465, 207Sánchez, S. F., Kennicutt, R. C., de Paz, A. G., & et al. 2012a,A&A, 538, 8Sánchez, S. F., Rosales-Ortega, F. F., Marino, R. A., et al. 2012b, A&A, 546, 2Sánchez-Blázquez, P., Peletier, R. F., Jiménez-Vicente, J., et al. 2006, MNRAS,

371, 703Schiminovich, D., Wyder, T. K., Martin, D. C., & et al. 2007, ApJS, 173, 315Seaquist, E. R., Bell, M. B., & Bignell, R. C. 1985, ApJ, 294, 546Sharp, R. G. & Bland-Hawthorn, J. 2010, ApJ, 711, 818Simien, F. & de Vaucouleurs, G. 1986, ApJ, 302, 564Smith, B. J., Giroux, M. L., Struck, C., & Hancock, M. 2010, AJ,139, 1212Smith, B. J., Struck, C., Hancock, M., et al. 2007, AJ, 133, 791Springel, V. 2005, MNRAS, 364, 1105Springel, V. & Hernquist, L. 2003, MNRAS, 339, 289Springel, V., Matteo, T. D., & Hernquist, L. 2005, MNRAS, 361, 776Stasinska, G., Fernandes, R. C., Mateus, A., Sodré, L., & Asari, N.V. 2006,

MNRAS, 371, 972Stockton, A. 1974, ApJ, 187, 219Teyssier, R., Chapon, D., & Bournaud, F. 2010, ApJ, 720, L149Thomas, D., Maraston, C., & Bender, R. 2003, MNRAS, 339, 897

Toomre, A. 1977, Evolution of Galaxies and Stellar Populations, 401

Toomre, A. & Toomre, J. 1972, ApJ, 178, 623

Valdes, F., Gupta, R., Rose, J. A., Singh, H. P., & Bell, D. J. 2004, ApJS, 152,251

Vazdekis, A., Sánchez-Blázquez, P., Falcón-Barroso, J., et al. 2010, MNRAS,404, 1639

Veilleux, S., Cecil, G., & Bland-Hawthorn, J. 2005, ARA&A, 43, 769

Vorontsov-Vel’Iaminov, B. A. 1958, Soviet Astronomy, 2, 805

Weiner, B. J., Coil, A. L., Prochaska, J. X., et al. 2009, ApJ,692, 187

Wild, V., Groves, B., Heckman, T., et al. 2011, MNRAS, 410, 1593

Wild, V., Walcher, C. J., Johansson, P. H., et al. 2009, MNRAS, 395, 144

Williams, M. J., Zamojski, M. A., Bureau, M., et al. 2011, MNRAS, 414, 2163

Woods, D. F., Geller, M. J., Kurtz, M. J., et al. 2010, AJ, 139,1857

Yun, M. S. & Hibbard, J. E. 2001, ApJ, 550, 104


1 School of Physics and Astronomy, University of St Andrews, NorthHaugh, St Andrews, KY16 9SS, U.K. (SUPA)

2 Institute for Astronomy, University of Edinburgh, Royal Observa-tory, Blackford Hill, Edinburgh, EH9 3HJ, U.K. (SUPA)

3 Departamento de Física Teórica, Universidad Autónoma de Madrid,28049 Madrid, Spain

4 Instituto Nacional de Astrofísica, Óptica y Electrónica, Luis E. Erro1, 72840 Tonantzintla, Puebla, Mexico

5 Instituto de Astrofísica de Canarias (IAC), E-38205 La Laguna,Tenerife, Spain

6 Depto. Astrofísica, Universidad de La Laguna (ULL), E-38206 LaLaguna, Tenerife, Spain

7 Instituto de Astrofísica de Andalucía (CSIC), C/Camino Bajo deHuétor, 50, 18008 Granada, Spain

8 Leibniz-Institut für Astrophysik Potsdam (AIP), An der Sternwarte16, D-14482 Potsdam, Germany

9 European Southern Observatory (ESO), Karl-Schwarzschild-Str. 2,D-85748 Garching b. Muenchen, Germany

10 Astronomisches Rechen-Institut, Zentrum für Astronomie der Uni-versität Heidelberg, Mönchhofstr. 12 - 14, D-69120 Heidelberg,Germany

11 INAF – Osservatorio Astrofisico di Arcetri, Largo Enrico Fermi 5,50125 Firenze, Italy

12 Dark Cosmology Centre, Niels Bohr Institute, University of Copen-hagen, Juliane Mariesvej 30, 2100 Copenhagen, Denmark

13 Department of Physics, University of Helsinki, Gustaf Hällströminkatu 2a, FI-00014 Helsinki, Finland

14 Max Planck Institute for Astronomy, Königstuhl 17, 69117 Heidel-berg, Germany

15 GEPI Observatoire de Paris, CNRS, Université Paris Diderot, PlaceJules Janssen, 92190 Meudon, France

16 Departamento de Fisica, Universidade Federal de Santa Catarina,Brazil

17 Centro Astronómico Hispano Alemán, C/ Jesús Durbán Remón 2-2,04004 Almería, Spain

18 Astronomical Institute of the Ruhr-University Bochum, Univer-sitätsstr. 150, 44580 Bochum, Germany

19 RUB Research Department ’Plasmas with Complex Interactions’,Universitätsstr. 150, 44580 Bochum, Germany

20 Institut d’Astrophysique de Paris, CNRS UMR7095, UniversitéPierre et Marie Curie, 98bis Bd Arago, 75014 Paris, France

21 Institute of Astronomy, University of Cambridge, Madingly Road,Cambridge, CB3 0HA, UK

22 Instituto de Astronomía,Universidad Nacional Autonóma de Mex-ico, A.P. 70-264, 04510, México,D.F.

23 Instituto de Cosmologia, Relatividade e Astrofísica – ICRA, Cen-tro Brasileiro de Pesquisas Físicas, Rua Dr.Xavier Sigaud 150, CEP22290-180, Rio de Janeiro, RJ, Brazil

24 University of Vienna, Türkenschanzstrasse 17, 1180, Vienna, Aus-tria

25 Instituto de Fisica de Cantabria, CSIC-UC, Avenida de los Castross/n, 39006 Santander, Spain

26 Sydney Institute for Astronomy, School of Physics A28, The Uni-versity of Sydney, NSW 2006, Australia

27 Departamento de Astrofísica y CC. de la Atmósfera, UniversidadComplutense de Madrid,E-28040, Madrid, Spain

28 Departamento de Física Teórica y del Cosmos, Universidad deGranada, Spain

29 Centro de Astrofísica and Faculdade de Ciências, Universidade doPorto, Rua das Estrelas, 4150-762 Porto, Portugal

30 CEI Campus Moncloa, UCM-UPM, Departamento de Astrofísica yCC. de la Atmósfera, Facultad de CC. Físicas, Universidad Com-

plutense de Madrid, Avda. Complutense s/n, 28040 Madrid, Spain

31 Landessternwarte, Zentrum für Astronomie der Universität Heidel-berg (ZAH), Königstuhl 12, D-69117 Heidelberg, Germany

32 Australian Astronomical Observatory, PO Box 915, North Ryde,NSW 1670, Australia

33 Department of Physics and Astronomy, Macquarie University, NSW2109, Australia

Appendix A: Pre- and post-shock gas density

Assuming pressure equilibrium between the post-shock gas andthe gas that has cooled sufficiently to emit [Sii], the CALIFAobservations can be used to measure the pre-shock density (n1)in the following way. The post-shock temperature (T2) is givenby

T2[K] = 1.38× 105

(

vs[km/s]100

)2

(A.1)

for a fully ionised precursor (Dopita & Sutherland 2003). Thetemperature of the gas emitting [Sii] is T3 ∼8000 K, and thedensity of this gas is measured from the[Sii] line ratio (n3 = ne).For a fast shock the post-shock density is four times the pre-shock density (n2 = 4n1). The pre-shock gas density is thengiven by:

n1[cm−3] =n3T3

4T2= 0.12

ne[cm−3]100

(

350vs[km/s]

)2

(A.2)

From this, we can estimate the thermal pressure of the cloudswhere the [Sii] lines are produced:

Pcloud = ρ1v2s = n1mpµv

2s (A.3)

whereρ1 is the mass density of the medium into which the shockis propagating,vs is the shock velocity,mp is the proton mass,µ = 1.36 accounts for an assumed 10% Helium fraction. Thisgives a relation between thermal pressure and electron density

Pcloud = 3.3× 10−11ne[cm−3]100

Nm−2 = 3.3× 10−12nedynes cm−2

(A.4)

close to the value given by HAM90 of 4× 10−12nedynes cm−2

based on shock models by Shull & McKay (1985).The bolometric luminosity per unit area of a shock can be in-

dependently estimated fromLshock= 0.5ρv3s , and this provides aconsistency check on the shock speed estimated from the modelfit to the observed line ratios. Summing over the full area ofthe two bicones, which we estimate from the CALIFA maps tocover 1/4 of a sphere, we find a bolometric energy loss rate ofEshock = 3.2 × 1043erg s−1 for vS = 350 km s−1. This is only afactor of∼2 smaller than the estimated bolometric luminosity ofthe bicones, which lies well within the errors of these calcula-tions, and supports the argument that 350km s−1 fast shocks arethe most likely candidate for causing the ionisation of the gas inthe cones.

Appendix B: Simulation methodology

The details of the simulation are presented in Johansson et al.(2009b) and Johansson et al. (2009a), here we provide a briefsummary of the relevant details. The simulations were per-formed using the entropy conserving TreeSPH-code GADGET-2



(Springel 2005) which includes radiative cooling for a primor-dial mixture of hydrogen and helium together with a spatiallyuniform time-independent local UV background.

The assumed dark matter profile has a significant impacton the evolution of the merger. Here we use an “NFW-like”profile, i.e. an analytical Hernquist (1990) dark matter profile,with a concentration parameterc = 9, matched to the empiri-cal Navarro et al. (1996) profile as described in Springel et al.(2005). Disks have exponential profiles, with a total baryonicdisk mass ofMd = md Mvir, wheremd = 0.041 andMvir is the to-tal virial mass. The disks are composed of stars and gas, withafractional gas content offgas = Mgas,d/(Mgas,d + M∗,d) = 0.2.The stellar bulge(s) have profiles closely approximating a deVaucouleurs law (Hernquist 1990) and a stellar mass ofM∗,b =13 Md = 0.27M∗,d i.e. close to the observed bulge-to-disk ratio forNGC 4676B ofM∗,b/M∗,d ∼0.25. All stellar and gas particles areembedded in a dark matter halo.

The galaxies contain 240,000 disk particles, 60,000 gas par-ticles (i.e. disk gas fraction of 20%), and 400,000 dark matterparticles. NGC 4676B additionally has 100,000 bulge particles,the bulge is omitted from NGC 4676A to match the observations.This leads to a baryonic mass ratio of 1.3, close to that observedin the real Mice galaxies. The baryonic and dark matter particleshave masses of 1.8× 105M⊙ and 3.2× 106M⊙ respectively.

The initial orbital parameters and time of observation(180 Myr following first passage8) were taken from Barnes(2004). The viewing angle was chosen to provide the best matchby eye to the CALIFA spatial and velocity maps, using the pack-age SDvision (Pomarède et al. 2008)9.

We follow the method of Wild et al. (2009) to assign ini-tial smooth star formation histories to the disk and bulge stars.For bulge stars, we assume a single formation epoch at a look-back time of 13.4 Gyr (the age of the Universe at the redshift ofthe Mice). The disk stars have close to constant star formationrates, with a slight exponential increase to earlier times to ensureconsistency between the current mass and previous star forma-tion history. The simulation then provides the star formation his-tory up until the point of observation, and we extract the spectralenergy distribution of each particle using the spectral synthesismodels of Bruzual & Charlot (2003). Yields are not tracked inthese simulations, so metallicity is kept fixed at solar in agree-ment with the measured metallicity of the galaxies (Section6.2).Similarly, attenuation of the starlight by dust is not included dueto the many assumptions required, and we compare to observedquantities that have been dust attenuation corrected.

8 Snapshots are extracted every 20 Myr.9 http://irfu.cea.fr/Projets/COAST/visu.htm


the mice at play in the califa surveyvw8/resources/mice_v8_astroph.pdf ·...

Documents